Tamara Laine: I think that it is such a revolutionary technology and I think that it has really been transformative and I do love the technology, but love and trust are two different things.
James Carman: Yeah, that's fair. Welcome to the Forward Slash Podcast, Welcome to the Forward Slash Podcast, where we lean into the future of IT by inviting fellow thought leaders, innovators, and problem solvers to slash through its complexity. I am your host, James Carmen, and today we're talking to Tamara Lane. Tamra Lane spent years as an investigative journalist and won an Emmy digging into ethical AI. She was also part of the 2012 UN campaign to ban autonomous weapons, standing alongside Nobel laureates making the case that some decisions should never be handed to a machine. She now builds autonomous systems for a living. Tamara is the founder and CEO of Empower, the first policy bound agentic infrastructure for lending. Tamara, welcome to the show.
Tamara Laine: Thank you for having me. Quite an introduction. I should say
James Carman: Yeah.
Tamara Laine: I was I was j merely assisting the UN campaign that the Nobel laureates were doing all the hard work. I was very
James Carman: See, I was gonna ask, I thought you were in the fighter jet that was going against the robot fighter jets to like pro like you were like you were the maverick of the yeah.
Tamara Laine: When I was an investigative journalist, I did do a air to air refueling mission, which was incredible. Yes.
James Carman: That's gotta be cool. Okay. All right, now we gotta talk about this. So all right, how
Tamara Laine: Yes.
James Carman: did it happen? How how how did you find yourself in this this c this scenario where you would need to do such a thing?
Tamara Laine: it was a time I if you if everyone remembers or maybe younger people won't remember this, but back in like 2012, 13, 14, a lot of gear, weapons, were being kind of sold or auctioned or distributed to police forces.
James Carman: Mm-hmm.
Tamara Laine: yep, so you remember all of that? A lot and there was
James Carman: yeah.
Tamara Laine: It was very controversial and so I covered that a lot. And part of covering that I got into a lot of conversations with the US military and was invited to do a few things with them that were really incredible to see firsthand, just really the skills, the training, everything that goes into some of the US Air Force missions as well as the military missions. it it is it is quite incredible. So kind of putting aside the the in the US controversy of U US police forces having military weaponry, white right?
James Carman: Mm-hmm.
Tamara Laine: Looking at how the military operates and and that was really incredible firsthand to be up there in the air. I also did some MRAP training and it was quite incredible. Yeah.
James Carman: That's awesome. That's so cool. All right. So you've you've had a really, really interesting background. I'm so happy to have you on the show because this is you have such a such a cool background. So you did investigative journalism for a while. You're digging into all sorts of stuff. And you do this story on ethical AI at some point. And that was that the trigger for you to kind of convert to to building these autonomous systems yourself, or did this happen sometime later? How did this journey happen?
Tamara Laine: The journey, so one of the media outlets that I was a part of helping launch had to do with impact investing. And what I was doing in particular was talking to founders and really looking at the startup scene. And that's where I started to fall in love with building technology, building startups, seeing how you could change people's lives. I I was working kind of how do I help people from the journalism side to how do you you as a company corporation change people's lives. And so I really became interested in that. And so when I left journalism, I I wanted to go into that zero to one phase. I I thought that it was incredible and had so much potential. but my life really collided with AI when I did this story on activist investors fighting for a code of ethics in Amazon and Northrop Grumman's technology AI and their AI development.
James Carman: Mm-hmm.
Tamara Laine: What I was doing was I was talking with ethicists and and technology drivers and and you know investors. I was talking to the whole spectrum of people who are involved in this one particular area. And I I found it just absolutely incredible. And that's where I really started to dig in and kind of continue down that path of AI and how it was going to interact with our with our world.
James Carman: That's that's just a fascinating story. The most people get the get the bug, like the the technology bug kind of early, but you you kinda had a different journey. I love it. That's awesome. and this this was, as your bio said, an Emmy winning invest investigative
Tamara Laine: Yeah.
James Carman: journalism. What what how I don't know how does what kind of Emmy is that? Like what what is the the official category?
Tamara Laine: It's a
James Carman: How yeah.
Tamara Laine: it's an Emmy. It was my goodness, I'm gonna forget the category. It was it was like short form documentary. I have to look back at it. I'm forgetting the the category that it was in, but it was it was a longer it was a longer form story. It wasn't a full documentary. So it was about
James Carman: Gotcha.
Tamara Laine: seven minutes. so it was like it was a special expose, but I just off the top of my head I'm now forgetting the actual category. If I I have it here. Do you want me to grab it?
James Carman: Yes. Well we want to know about this. This is cool stuff, yeah.
Tamara Laine: Okay. Sorry. Here it is. eyes on you, keeping
James Carman: Gosh that's so cool.
Tamara Laine: recognition companies in short short form content. So technology short form content up to ten minutes. That was it.
James Carman: All right, so this is so cool. She's got this I you can't see it, but she's got the statue. She's holding it in her hand. Like that. This is I
Tamara Laine: Yeah.
James Carman: I don't know if I know a lot of people. I think we've had guests that we've had one other guest that had an Emmy. I think it was a daytime Emmy for something related to public television or something like that. And then we've also had someone who had a Grammy. one of our folks, one of our guests had a Grammy. So we're we're still looking for that Oscar and the Tony. We gotta get we're we're going for the egot, but but That's that's really cool. now how did and I gotta I gotta geek out on this. So did you like go to the Emmys did like to receive the award or was this one of those like well it's short from life mail it take? How did this work?
Tamara Laine: No, that well, this was such a bummer. it was the it was COVID. So I sat on the couch, dressed up with my family, and my daughter thinks that she won the award. She was very proud of herself. Yeah. Such a young age, so impressive.
James Carman: well that's good. Good for her. At such a young age, even. That's impressive. Yeah. Yeah. So they did they like film you like when the when the announcement happened and everything? Like did you did you have like your phone?
Tamara Laine: No, it was just so sad. No. yeah, but that's okay. I still you know, the accomplishment is what it is. Not the not
James Carman: Yeah.
Tamara Laine: not the evening. And I went out for drinks with my producers later and it was a a good time.
James Carman: That's really cool. That that's that is such a cool story. All right. So the the ban killer robots is the campaign, right? That was the
Tamara Laine: Yeah, the campaign to ban climb.
James Carman: Okay, tell me about that. Like what what what all what was that all about? I want to hear more about this.
Tamara Laine: Yeah, well that was again so very early in my career, and and what it was was it was several Nobel Peace laureates and and people, you know, basically academics that had been studying this and as well as people in the military who kind of came together, or former former military came together and saw that this movement that technology was becoming incredibly smart, right? And what they wanted to do was get the government to put in regulate like controls, control levers around autonomous weaponry, right?
James Carman: Mm-hmm.
Tamara Laine: And so they were going all over talking at you know different government events. when I when I was volunteering in this and and working with part of the team, we went to the UN. And we sat in a forum on this and discussed, you know, its it its significance. And this was back in 2012. And so kind of where I find irony is back then when we were talking about it, people were like, this is not gonna be a huge deal, right? Like, this
James Carman: Mm-hmm. Mm-hmm.
Tamara Laine: isn't this isn't my problem today, right? They ever was just like, you know, like w no one's gonna let autonomous weapons just roam around. And they were already, right? Like there was autonomous, guns actually on like border control. you know, our our drones of that time could be autonomous, but we just didn't let them. so it was really a a significant conversation to to happen. but I just I remember sitting in these rooms and we were having these discussions and just kind of people being like, like that's not gonna happen. And now if we're, you know, we're 10, 12 years, 14 years later, it's not like Everybody, if you pull people in the US, they're like, yeah, right? Like, yeah, we've seen we've seen the robots walking around. Like I'm in the Boston
James Carman: Mm-hmm.
Tamara Laine: area. Like at our children's museum, the Boston Dynamics,
James Carman: yeah.
Tamara Laine: you know, robot that can walk over any terrain across the world. It's just the it's like moving around in the children's museum, right? Like it's there. So like we're watching what's what's that championship in China right now of the
James Carman: Saw that, yeah.
Tamara Laine: the the robots, I for what's the name of it there? The two legged yeah, the two it's like
James Carman: It was like a Robot Olympics kind of thing, wasn't it? Yeah.
Tamara Laine: two legged robots and just to see and like they can run faster than humans. It's in it's wild. It is wild how far we like the technological technological advancements in the last twelve years, which is just it boggles my mind. But yeah, so then and I remember again I was super young back in this when I was working on this campaign. I guess I I wasn't super young, I was super young in my in my understanding of this and my work in it, right? I would say I was pretty, pretty junior. and I was sitting at the UN in the round in in the circle. And because they knew I was a journalist, the team asked me to ask the the ambassadors at the table to like raise my hand and be the one who brought this up. And so I I
James Carman: Nice.
Tamara Laine: brought it up to like the ambassador of Egypt. like raised my hand and and started the conversation on, hey, what are we doing about autonomous weapons and, you know, regulate I forget the question, but I I do remember the the s the situation because I was just like, you know, it was I was also very early in my journalism career. So I had sat across heads of states before, but it's, you know, when you start asking those questions young in your career, you're You get a lot better at it when you get when when you get your your C Lane.
James Carman: How was that received with with like, you know, just this this person, a journalist asking these heads of how was that received? Were they like who are you to ask that? Or were they were they like just answer?
Tamara Laine: Well, no, because you're sitting around the table, so they just answer it, right? Like it's not
James Carman: Okay.
Tamara Laine: yeah, they didn't think I don't think there were people thought much of it other than just like you've got to seat at the table, so you you've kind of earned the right to ask questions. But yeah, but inside
James Carman: Nice.
Tamara Laine: I was like, Whoa, don't screw this up.
James Carman: Yeah. my God. Wow. Man. All right. We we could just talk all day about this stuff. But I but you're here to talk about AI and and what you're doing now. And what you're doing is what everybody's talking about. Trying to build safe, guardrailed,
Tamara Laine: trusted.
James Carman: contained, trusted, all of the things, right? give us those warm, fuzzy feelings about our AI agents. That's what you're doing. That's what you are doing with empower. That's that's what the kind of is that the mission of the of the company is. you know, kind of building safe and
Tamara Laine: Yeah, I mean, well, we're building in highly regulated industries, right? So we're we
James Carman: Mm-hmm.
Tamara Laine: are the policy engine for for orchestration and underwriting right now is our is our niche and know that we'll expand. But what we are what we are doing is we are building highly guarded, you know, regulated agents that are in an orchestration a layer that can do the work. for right for sorry lenders. I was gonna say regulators. for lenders. And that's real, really challenging because I don't trust AI. Right.
James Carman: I don't think anybody does right now. That's that's the problem. Yeah.
Tamara Laine: So yeah, I've got a really incredible team and we're creating a product so that we can really you know the the underlying goal is improve access to capital for for everyone, right? So improve the lending process for everyone. And so our agents, what we're doing is we're building the policy engine and the infrastructure so that agents can do all the work and orchestration for a highly complicated, highly regulated task. And so in order to do that, you do. You have to, you have to build, you know, it has to be transparent, auditable, you have to abide by all these rules. The guardrails have to be very secure, right? Like it's it It's not for the faint of heart.
James Carman: Yeah. Yeah. I wanna it's interesting because a lot of people are are talking about this stuff in a in a very kind of hand wavy motion right now. There's like, well, we gotta put guardrails on these things. Yeah. So
Tamara Laine: Airbase creating agents. Yeah.
James Carman: we're gonna create agents. We we got we know we gotta put guardrails on this stuff. What yeah, we we're gonna make it safe. It's very surface level conversation, but you're getting down into the meat of it. So it'd be interesting to hear your perspective on kind of what are some of the the key aspects of of of this workflow. Like, you know, key components of of being able to make a good decision that is auditable, that is traceable, that you can you can back up and you can say, Yeah, this is why you know, all of all of the things, right? So may me can we break it down maybe into some componentry here? Just some low level like you you gotta get this part right, this part right. Can we do that?
Tamara Laine: my goodness. Let me let me think of how to how to answer that. because I we're in the we're in the weeds right now, right? So like
James Carman: Yeah, yeah, yeah.
Tamara Laine: so ask me that question again just so that I make sure I'm I'm hitting it correctly.
James Carman: Well, what are what are some of the well first let's start with some of the failure modes, maybe. Like, so okay, an agent in your domain is is trying to make a decision based on is is the idea basically someone can or cannot borrow money. Is that the idea?
Tamara Laine: Yes.
James Carman: In a very simplistic Yeah. Is that what you're saying? What w that's the domain where
Tamara Laine: Yeah.
James Carman: where you work?
Tamara Laine: Yes, exactly. Can or can so our agents are helping parse through the data to give to an underwriter to make a very fast decision on whether or not someone ki in within the institution's policy qualifies for a loan. So
James Carman: Okay.
Tamara Laine: we've got a lot of working parts on that.
James Carman: Yeah. I know there's and there's there's a lot of regulation. I know is it isn't it Colorado has got some stuff that's activating. Was it January? They pushed it back to January of twenty seven. Is that right when they're activating? Yeah.
Tamara Laine: Yeah. And and you see the same thing in the EU. Like we're actually really following
James Carman: Mm-hmm.
Tamara Laine: the EU stuff because that usually is a little bit more progressive than any of the state stuff.
James Carman: Yep, yep.
Tamara Laine: and stuff is coming out of the Treasury Department as well. But yeah, I lending and underwriting is considering the most high risk task
James Carman: Mm-hmm.
Tamara Laine: for AI. And so what you have to parse through with that is to make sure that no and and actually I just firmly believe AI should never make the decision. So what AI needs to be doing is the the grunt work, right? It needs to be gathering information, going back and forth, doing a lot of the communication with the borrower, like missing documents, standardizing documents, right? Like when you're getting documents from multiple dis different sources, what is the source of truth? Where, how do you how do you solve conflicts and information and with what with what trust value?
James Carman: Gotcha.
Tamara Laine: And working in a highly regulated, a lot of people, I think this is my my controversial take, but a lot of people do not are pushing for less regulation in AI. I find that having regulations, firm regulations, right? Standards that you have to operate off of give us a lot more guardrails to keep things safe, right? And
James Carman: Mm-hmm. Yeah.
Tamara Laine: that way, that way we have to operate in a safe way. We have, you know, like there isn't wiggle room for you know, bias. We we can't. We just can't, right? We can't shift something that is biased, where a lot of other industries can, right? even in security, you can do that, right? Like it's just it's I think it's really good. And having something that's firm that we know we have to operate is way better than having some ambiguous rules. That we don't know which way the wind is gonna blow and how it's gonna affect us, right? And I I say that's the killer of a startup. if a startup has ambiguous regulations that they're trying to figure out, then funding starts to dry up. You know, their product can be very vulnerable if the wind blows the wrong direction. So, like I enjoy working in a highly regulated area.
James Carman: Okay. Yeah, I think I I I love this analogy. My wife and I went to a parenting seminar. It's I'll circle it back to make it relevant, but it's it was a parenting seminar and the guy told a story about a study they did of of these kids. And what they did was they would they would have these children go out on a playground and there was no fence around the playground. And when there was no fence around the playground, the kids would kind of they would hover near the building of the school and they wouldn't they wouldn't kind of go out and play on the the different pieces of equipment or whatever within the playground and they kind of hover near the school and they wouldn't, you know, venture out. Then they put a fence around the playground and the kids were flourishing and they were playing around and they were jumping all over, going on the monkey bars and all that. So that it kind of lends itself to that idea of what you're talking about. When we have those guardrails and those boundaries, it does allow us to know here's our playground. Here's where we can play and we can play freely and we know we're safe. I agree with you. I think the big thing that people are arguing about right now is Who makes up the rules? Because you as you said, like you don't want it to be just because an election happened or whatever, you know, the the the the winds change, somebody gets in there and they want to, you know, rip the rug out of it from everybody. Like we we don't want that. We want we want a nice stable environment with good rules that that we can all play with with the same rules. I think that's the the big argument right now. But yeah. yeah. Cause
Tamara Laine: Exactly.
James Carman: that is the scary thing, and that's those are the things that have been happening to people is. They get rug pulled with, you know, a regulatory rug pull where they they have a business model and then someone just passes a law and wow, my business is now gone. I there's nothing I can do about it. That's kind of the scary thing. we're we're talking about borrowing money. I know I know this this is absolutely one of the things. As you said, it's you you don't want it to make the decision. And it from a regulatory standpoint, I do think the wins are already there of of saying, like, you're not, you're not gonna make decisions about, you know. Does this person deserve to get healthcare? Are they are they gonna get a loan for a home or w get a job? There's they've already carved out like these are some of these key life-changing decisions for humans that we're just not gonna allow AI to do. so that's good. Yeah.
Tamara Laine: Yeah. Yes.
James Carman: I and I love what the way you kind of phrase that. Tell me more about this, the grunt work thing, because I think that's that's a that's a huge pattern in this in this agentic world is. Go do all the work and gather up the body of work and put it in front of a human being and let them make the ultimate decision. So say so say more about that. How how does that work?
Tamara Laine: Yeah, so the idea is to take take away the busy work, right? The the back and forth. I mean, so for us, we're pulling data sources from hundreds of different we're pulling data from hundreds of different data sources, right? and for for you know, for lenders and kind of caveat like, you know, every lender is different, has different capacity and and different IT teams and all of that, right? but kind of inevitably that's the part that slows down every process. So if you look at
James Carman: Mm-hmm. Mm-hmm.
Tamara Laine: mortgages, like mortgages can take between 30 and 45 days. And it's not because it actually takes that long. It's because things are passing through so many different systems that it just drags. It's called like exception drag. And
James Carman: Mm-hmm. Absolutely.
Tamara Laine: so Asians can just do that more efficiently, right? They can pull the information, they can parse through it, they can find out what's missing, they can, they can get in touch with a borrower. They can get the information from the borrower. They can file things in the right spots. Like agents can just do this in a more efficient way with the right rules and bounce put on them. And so that's what we're doing. I mean, we I I got up and I gave a keynote a while ago and I said, you know, AI should be your your trusted junior analyst. And that's how you look at it.
James Carman: Yeah. I like it.
Tamara Laine: Because we don't trust AI.
James Carman: Yeah. I yeah, I don't think it's earned it yet. So I mean it's it's it's it's true. And I don't know that it ever should for some key decisions as as we're we're carving
Tamara Laine: Yeah.
James Carman: out. I don't I don't think it's ever should should earn to it's the right to make those decisions that are truly that impactful on a human being's life.
Tamara Laine: It really I mean, like I I said that twice and joked around. I think that it is such a revolutionary technology and I think that it has really and I ha I I do love the technology, but love and trust are two different things.
James Carman: Yeah, that's fair. I think that's fair. Well, what about? So in this process, I guess the one failure mode that I can think of off just off the top of my head, and I'm sure there are others, is the agent goes through gathering information, the the body of work and says, Hey, underwriter, here's the the case file for this thing we're doing, right? say yes or no to this, and you you make the the ultimate decision. Well, what if it doesn't gather complete data? That would be kind of the one failure mode I can think of, is that it didn't give you all the information that it should have or could have for you to make a more informed decision as an underrated.
Tamara Laine: Yeah, so you have to put in all sorts of evaluators throughout the whole process, right? Like,
James Carman: Mm. Okay.
Tamara Laine: like is is this trusted on a scale you know, of one to one hundred, what value do we give this information coming through so that when it does get to the final underwriting package, you have not only like the underwriting package package, but the trust score on multiple different sources of what you're evaluating
James Carman: Mm-hmm.
Tamara Laine: against. So that you know, like again, if you're an underwriter and you're getting something that like, okay, you say it's ninety percent, you can go, all right, well, I've been using this for a while, ninety percent is usually pretty good, but let's just say it's like eighty percent, and you're like, well, I'm gonna go double check that. So that you can
James Carman: Mm. Okay.
Tamara Laine: get really, you know, a better picture and and use your own judgment of whether or not you need to go back. And is it an area that is highly important to the organization or is it not, right? Like you can you can make your own judgment on that. but it's also really important. I you asked a question earlier, it's something that's important in agentic workflows, especially with with decision and with highly sensitive information is making sure your agents don't have the keys to everything.
James Carman: Mm-hmm. Yeah.
Tamara Laine: So how you can trust things more is if agents, you know, like every agent has its own little domain and then passes information
James Carman: Okay.
Tamara Laine: along and no one, no one no one agent can operate the whole thing.
James Carman: Okay. Yeah, I've seen a lot of patterns like that where they're these multiagentic systems where, you know, and people are setting them up, like, I run my company with, you know, all these little QT named agents and all that. Yeah. Yeah.
Tamara Laine: You see it on Instagram. This is about what was it? This is my Jarvis.
James Carman: Wake up, Jarvis, Daddy's home, you know, whatever, right? Like the
Tamara Laine: Yeah.
James Carman: the I I've seen that a gazillion times. but I I do think there there is a pattern. And I think there's there there there are studies. I don't have the data in front of me, but there are studies about, you know When you when you do have these kind of almost adversarial agents working together so that they own a domain and they have their own goals that that they want to do and they kind of balance, it's like a checks and balances kind of system. And as you said, one singular agent doesn't have all of the tools available to it that it can just go out and do everything. You kind of have to ask them to ask others and they're they're guardians of their own domain. I think that's that's yeah, yeah. Yeah.
Tamara Laine: 'Cause that's where the hallucinations really start to happen, right? When it when it's too much information for an agent and they're like, you're yeah, I mean, I'm sure everyone anyone who's used ChatGPT or CLOD or any of the two has been like, Are you sure about that? And they're like, you you're right. You were good to double check that.
James Carman: You're right to ask that.
Tamara Laine: Yeah. Your your instincts were spot on.
James Carman: And it's it's both right and wrong, but it's wrong for the right reason. Like, like what?
Tamara Laine: Yeah.
James Carman: What did you just say to me? Like, and that's the reason that matters. Like, don't patronize me, Claude.
Tamara Laine: Yeah.
James Carman: Okay. You know, I got stuff to do. I I literally was doing this just this morning, so it's hitting home for me. All right.
Tamara Laine: Yeah.
James Carman: So we're we're we're using multiple agents, and I think that one of the keys, as you're saying, is is that these agents maintain their own context so they don't bring If you if you use the same model, let's say it's Opus four or Opus five in in Anthropic, and you give it all the exact same context every single time. If everybody's just throwing the exact same context and the exact same data at it, it's going to make roughly the same decision. So that the important part there is when you partition these agents out, they manage their own context. They don't bring everything, all of the same information to bear to the to the LLM to say, help me make a decision. So That that's key because it it can water it down. When you when you pollute one call to an LLM with all of this information, it's gonna forget half of it by the time it gets around to generating stuff to talk to you. so yeah, I I I okay. So that's an interesting concept. Multiple agents, different contexts, controlling that. I love it. I wanted to come back to the word, the T-word that you mentioned, the trust and and and the data that you are bringing in, because that's that's a key. That's a key failure mode that people are seeing now. And and and I think that's going to be like the the software engineer of the future is around like understanding, okay, this data is coming in, but that was generated by an LLM. So I don't know that my trust score doesn't go high on that thing. But you know, this comes from what the human said. The human being actually said this. So I trust that more. Or this can't comes you know from some factual database. I can trust that more. But this was. This was coming in from an an L L generating information. So it's it's that prompt injection idea. Like I I I don't wanna allow for someone to creatively create a prompt that generates information that now I just put a trust score on that just muddies everything up. Is that are you seeing that? Is that are those some of the problems you're working around as well?
Tamara Laine: Well, our data, so the data that we're bringing in is maybe different than other data sources that other agents might be working off of.
James Carman: Mm-hmm.
Tamara Laine: because like we're we're bringing in pretty we're not bringing in data from agents yet, right? So I know Yeah, I yeah,
James Carman: Okay. They're not generating and they're not giving their your commentary. That's not their job.
Tamara Laine: I know the agent to agent discussion and I actually have been talking in in some of my startup groups, there's a really interesting thing that's happening, which is an agent to agent bank for agent to agent commerce, right? Which is
James Carman: Mm-hmm.
Tamara Laine: I like think about the things that we're now talking about, which are like, what? That would have been a conversation three years ago. but the Yeah, what a time to be like, but you know, it's a
James Carman: What a time to be alive, huh?
Tamara Laine: real I can't remember who I was reading was talking about this, but like just for general search now, right? Like if you were If you're someone who uses any of the the LLMs to research for like writing, right? or for commerce, if you are researching for like, I want to use a new deodorant.
James Carman: Mm-hmm.
Tamara Laine: what's what's really interesting is I forget the exact number, but it's it's kind of astounding. It's like 60% of the content that LLMs are scrolling, or it's like 60 or 70%. Is is content made by LLMs now because the actual sources where you would want to get your information from, like consumer reports, right? Does not allow scrolling on their page. So right. So they they're blocking. So like just like the New York Times or any of these other sources are blocking the scrolling, the legitimate sources that you really want to get your information are blocking the LLMs from scrolling and get and
James Carman: Mm-hmm.
Tamara Laine: retrieving information. So what you're actually doing now when you're searching is you're searching and getting information from a site that probably was created by someone using an LLM, meaning how do you trust that? so that's a really interesting, you know, thing
James Carman: Mm-hmm.
Tamara Laine: to think about that like it you kind of want to go back to Google.
James Carman: Do you remember? Yeah, I'm I'm giggling here because it just I don't know. Again, I I'm a movie guy and this reminds me. Do you remember you remember the movie? I think it was called Multiplicity with Michael Keaton, where he was like, he was just so busy, he's my gosh, if I had if I just had a clone of myself, well, I'll be darn, we can clone you. And they kept cloning, and then the clone would say, I'm too busy, I'm gonna clone myself. And by the like three generations deep of clones, the guy was like just a buffoon, right? He's like shaving his tongue, you know. I mean, it was that for me was silly stuff. But that's that's kind of what we're getting to. If if these L LMs are generating information that's gonna train LM so and it's going farther and farther away from ground truth, human knowledge, where's that gonna get
Tamara Laine: Yeah.
James Carman: us? Is it gonna get worse and worse? I d I don't know.
Tamara Laine: Yeah, it it really I mean that's why like when I do a lot of my writing, even though I do use several different LLNs to like, once I compile my thoughts and like do my rough draft to polish it almost like a copywriter, I do
James Carman: Mm-hmm.
Tamara Laine: a lot of the searching on my own from from like the websites that I know and trust because yeah, otherwise you're coming back with some blog from a law firm that created it using LL.
James Carman: Mm-hmm. Yeah.
Tamara Laine: Yeah.
James Carman: Yeah. It's it is so interesting that i even even when you do set up, because I have a bit of a a workflow that I work through with using agents to help me with writing and they they do research and adversarial work against themselves. So there'll there'll be a skeptic, right? But even through all that, it's still some garbage gets through every night. And you're right, you have to kinda you have to go ahead and check those things. But at least it does it does still streamline things. But that but you're right. We you do wanna go verify some of those assertions and and facts and quotes and stuff like that that it's asserting for sure.
Tamara Laine: Yeah. Well, and this goes, this is my advice to to younger people who are entering the workforce or who are, you know, really trying to get ahead. If you are not like if you're not double checking your sources and doing that work and doing that diligence, it it's so easy to spot. and you have to train yourself. Like again, I think we've we've talked we touched on this. AI is great for experts, right? It's really
James Carman: Mm. Mm-hmm.
Tamara Laine: good if you're an expert in your domain because
James Carman: Yes.
Tamara Laine: you are able to spot and to see through and use discernment on what's what is accurate, what is not, what doesn't feel right. Let me double check this. But if you're just going through several different prompts and then handing it over to someone else to look at, yeah, usually you can see through that right away.
James Carman: Yeah, and I think it's it has to do with the the confidence that it generates things with. So if you're not if you don't have that level of discernment and familiarity with the domain, the the LLM it's not like, Hey, I think this is true, but you know, maybe you might want to check. That's not how LLMs work. They're like, This is the facts and it's completely bogus, right? Yeah, it but it's gonna tell you confidently that that it, you know, a fact.
Tamara Laine: Yeah. I was doing some legal work, trying to I I ended up the the end of the story is I ended up just getting a lawyer, but I was using all the different tools, kind of going back and forth to do some some legal contracts. And I I said, Claude gives me anxiety doing that because it would come back a critical alert. Stop before moving on. Don't do anything until you do X, Y, and Z. And I was like, God, God, God, right? And then I would take it over to The other system and it would be like, no, according to this law, you're fine. but this is now an issue. And I was just like, I was going back and forth. I was like, this is not my domain. This is not. And I finally was just like, nope, this is why you do hire some people. This is this is why
James Carman: Yeah.
Tamara Laine: enterprise software will still be alive, right? Like this, all these things. It is because there are There are experts and there are people who do things really well and they do it well for a reason because they spend their time dedicated to it and let them do their good work.
James Carman: Absolutely. Absolutely. Yeah, I th the law thing is interesting. And you're and you're right, like the personality of the different models, you know, just based on their training is it they will give you interesting answers and then they'll disagree with one another based on, you know, just kind of their principled level thinking and how they're how they're trained to think.
Tamara Laine: Yeah, but Claude comes up and it's like red or green. Like it really does.
James Carman: Mm-hmm.
Tamara Laine: It's kinda it gives it gave me anxiety. I just like, I can't do this anymore, Claude Stop. You freaked me out.
James Carman: Now it you you said something that was interesting. So so so the whole idea of like checking your sources and all that, that probably comes natural to you. And you feel like that that journalism background makes you more adept at working with these these LLMs because you you have to do that as a journalist. I mean, you're not gonna publish a story if you you're not checking your facts. You can, you know, I there's There's ethical guidelines there. I mean, there's a you know, you you need to do that. Do you think that background is helping you in this this world of AI?
Tamara Laine: I think that that background helps me with a lot of the stuff that I'm doing. Yeah. Yeah, definitely a lot.
James Carman: Mm. True. Yeah. Yeah. It's interesting. We'll have to maybe have everybody, if you're gonna be a software engineer in the modern day, take take to a journalism class, you know, just there's there's a lot.
Tamara Laine: I honestly I think journalism journalism and law are two things that really translate well because it's you know how it's how do you think, how do you test, how do you, you know, determine validity, like all those things. And I think I think
James Carman: Mm.
Tamara Laine: those two really do a good job in creating a mindset to to work and deal with AI.
James Carman: I think it it's it comes down to we talked about this, I think, on the podcast in the past, but it's it's really about it's about critical thinking. And in in those jobs, that's one thing they they teach you. You have to be able to be a critical thinker to to do those jobs well as a lawyer or journalist and economist, you know, those sort of things. Okay. All right. So interesting, interesting stuff. And and you're right in the thick of it. It's it's it's again, what a time to be alive. Figuring out how we can trust these robots to make decisions for us or not, and how to put guardrails around them and how do you partition them correctly. I mean, we could we could sit here all day and talk about this. But we have to move on to our next segment of the show. Our next segment is what we call ship it or skip it. And this is our our take on hot or not. The idea is we bring up a topic, we talk about it, and we think, yeah, yeah, I like that. Let's ship that, or no, I don't like it, skip it. that's the gist of it. One that I think's interesting, What do you think about this notion of watermarking AI assisted work?
Tamara Laine: Skip it. am I allowed to elaborate at all? okay.
James Carman: Okay. No, absolutely. Yeah, yeah, yeah. The lightning rounds where we gotta get like terse. You know, we because they're act they're yeah, yeah, yeah. You can you can elaborate.
Tamara Laine: okay. Yeah. No, everybody is using AI to help them write. And beyond that, like what what is watermarked, right? So like if I if I do most of it and then AI comes and cleans it up, right? And it still has that AI hint to it. So that's watermarked. I think this is I think it's pretty lame. I think, you know, I get content on social media that is like deep fakes, right? I I get
James Carman: Mm-hmm.
Tamara Laine: wanting to identify some of that stuff, but from now, I mean, the genie is out of the bottle. People are using it to write, people are using it for websites, for graphics, for everything. And so I I don't think marking it is going to do anything other than like It was who I forget who it was, but just wrote the hedge fund guy, very very respected. I'm forgetting his name off to my mind. It's been there's a huge backlash on it. He wrote a wall a Wall Street Journal article, or Wall Street Journal op ed. And
James Carman: Mm-hmm.
Tamara Laine: and someone put it through the AI jet the AI test. And said it was, you 100% AI generated. And like the and the guy came back and said, Yeah, he goes, I do a lot of things in my day. And writing an op-ed is not one that's gonna take a lot of my, you know, my time. It's all my thoughts. I approved it. It is everything that I think. It, you know, it's my research. It's but I didn't sit down and handcraft every single word, right? and and that's what I think is important, right? Like, is it your is it your thoughts? Is it your your thinking? And like the Wall Street Journal was like, We've known this guy for, you know, decades. This is this is his thoughts. It doesn't matter if AI helped compile it and put it together. No.
James Carman: With you. I think early on when the the whole AI thing, and my kids are still like this. Like, gosh, dad, you use that AI stuff. That's garbage. You know, it's because the schools have scared them away from AI for some, because they don't want them cheating, right? So it's kind of a cop out. Anyway, early on, there was this mentality of like, AI slop, and you know, everything generated with AI is just garbage automatically. I thought we had gotten past that, but this watermarking thing, I think, is going to be a setback for us. And again, I not that I think everybody should be just throwing around, you know, just say, hey, write me an article and just start publishing things. No, as you said, do some research, have some ideas, develop a thought and and come up with something novel that's adding value to the world. If AI helps you write the article, okay, whatever, right? And you know, at least maybe it'll be better done than what you would have done by yourself, but you're you're able to express yourself. I I think it's a setback and I I I don't like it. I'm with I I'm trying to figure out like what is what is the real problem they're trying to solve. You said meant you know, kind of the social media, some of that deep fake stuff. I mean, that that stuff is like I it's some of it's it's just you just scroll past it these days. And it it's but it's a there's a lot out there. So I I don't know what problem they're really trying to solve with that. And I'm not a huge fan. So we'll we'll see. I don't I don't know.
Tamara Laine: Yeah, I mean, would I like less noise? Absolutely.
James Carman: Mm-hmm.
Tamara Laine: do I believe that like kids should learn how to write? I I I even now try and find time to do writing myself.
James Carman: Mm-hmm.
Tamara Laine: because it's a skill you don't want to lose, right? Like it is something that helps with your critical thinking, it helps with your brain development, it helps with all of that. Like you need to do it. you don't want to totally outsource it. We have to teach our kids how to do it well. You can and it goes back to like experts use AI well. If you're not a good writer and you use AI to write, it's going to be bad writing. If
James Carman: Mm-hmm. Yeah.
Tamara Laine: you're a good writer and you use AI to help perfect, it's gonna be really good. so yeah, so I think that there are things that we need to think about outside, like there's a more important conversation to be had instead of watermarking this stuff to like really instilling in our in our kids and and University is the importance of critical thinking and writing.
James Carman: Yeah, I I I'm trying to remember. I I did some research for a talk I gave not long ago. And this was this there was a study, and I'll have to probably should prepare better for these episodes. but they they they
Tamara Laine: I well.
James Carman: pointed out if they pointed out exactly what you're saying, that they gave they had writers use AI and there was a there was a benefit, and there were people who don't write, and there's so there's like three levels of people and the the findings were interesting. It was the the domain experts using it. about subject. They're not writers, but they're domain experts. It helped them the most because they had the discernment and all that stuff. A professional writer, the quality of work that the AI gave a professional writer, someone who knows how to write really well, is I'm recalling this off the top man. My my recollection was that that was kind of it was it was not great, right? But then the people who weren't domain experts and are not writers, mm. wasn't good at all. Right. So I think that but the biggest bump was the domain experts being able to exp help them to write better stuff. That was where the biggest bump was.
Tamara Laine: makes sense. I mean it's kind of the same how I feel with coding, right? Like
James Carman: Mm-hmm. Yeah.
Tamara Laine: there's vibe coding now can make a poor to mediocre coder 80% better, right?
James Carman: Mm-hmm.
Tamara Laine: Which like covers up a lot of a lot of non-domain expertise. but then
James Carman: Mm-hmm.
Tamara Laine: if you give it to someone who is a coder and is really good at doing it, you're gonna get something that's excellent.
James Carman: Yeah. Yeah. I I mean that was one of my early findings when starting to use this stuff is, you know, it helps me tremendously because I've been doing this a long time with you. Again, I think it you I love the word you use, the discernment. I I think that's a that's a key phrase and that's that's but that's earned. That's that's experiential. And the problem we're gonna have is If you're not doing the stuff that I did for 30 years, you're not gonna get that dismant. So it's kinda hard to build that dis that's gonna be the the the the thing we need to solve for is how do we get people the experience to get the discernment to be able to do that in the in the the future when they're not having to do the the hard thing. So that's that's gonna be the million dollar question or billion dollar problem, maybe trillion. We're we're talking crazy numbers now with AI. So it's probably trillion dollar question. All right. So the next ship just recently, Claude came out with this auto approval mode. And I guess the gist of it was what they found was, or anthropic, I should say, they they found that most of the time when they presented the prompt for the human in the loop to say, Hey, I want to go do this thing. Can I do that or not? The human would just rubber stamp it most of the time. It was just boom, boom, boom. Just go ahead and do it. Don't quit bugging me. So they've turned on this this auto mode where the they have an agent itself or some form of of yeah automated process that that makes the decision is Claude allowed to go do this thing that it wants to go do? what are your thoughts on that?
Tamara Laine: I guess the the question is what is the task, right? Like I
James Carman: Mm-hmm.
Tamara Laine: I think that it's a much broader question that probably deserves a lot more review before just turned
James Carman: Mm-hmm.
Tamara Laine: on because the answer is gr is completely different with different tasks, right? Is it a so I mean, I was gonna say is it a social media post, but I would still wanna be in I wouldn't want any anything automatic with anything that's public facing. Right. is it is it internal, is it external? Is it financial? Is it social media? Like why so many questions for me to just have a really generalized, yeah, bump it, you know, go ahead and automate this and and run with it. so I don't think that's an easy one for me to say ship it or skip it, because it's kinda like, I guess I guess I would say skip it because there's too many questions in my head.
James Carman: Too much. Yeah.
Tamara Laine: Yeah.
James Carman: It's interesting to me. I I so I would be hypocritical if I was to say completely skip it because for a long time now, and again, I'm doing, you know, research type work and and that sort of stuff when I'm doing coding these days. I'm not doing, I don't, I don't do a lot of client work these days, but most of my stuff is in R D. anyway, when I fire up Claude, I'm firing it up in YOLO mode, right? Like I don't, I don't want you asking me a bunch of questions. So that's what I've been doing for a long time. So I think I haven't really had any problems. Then I don't know. Maybe, maybe it's offloading all my information to some Chinese company or something. I maybe that's what's going on. I just don't know. but but it I haven't had any issues with it. So I think there's there is there is a there there. so it's incrementally better than what I'm doing and saying, hey, Claude, just don't ever ask me. It's it's incrementally better than that by like at least putting some level of judgment between YOLO and you know. I think it does get to the point where it's like, wait a minute, this this feels like this might be a little. I'm not super confident in saying yes to this. I'm gonna have to ask the human, you know, person to to get involved here. I think there's that aspect already built in. It hasn't asked me yet. I haven't seen a place where it's asked me, but I don't know. I I'm I'm skeptical, but I'm I am optimistic. And they they did do some research as far as far as like we we matched it up against when a human answers the question. Think there's especially Claude code, right? Because it asks to do some like bash commands, like stuff on your machine, like, hey, can I remove your root file system? Yeah, sure, no
Tamara Laine: It yeah.
James Carman: problem. Claude, yeah, go for it. I don't know what that RM minus FR means. I have no idea. but the agent that or whatever system they use, their auto approval thing did, and it was able to say, Yeah, probably don't want to do that. But the humans were, they're just rubber stamping the stuff where that, whereas the agent was actually triggering and and turning it off. So I think there is a there there. I'm with you though. I think when it comes to things that are like I'm not gonna have auto mode just shoop shipping emails out to my clients or posting stuff on my behalf to social media or buying
Tamara Laine: Yeah.
James Carman: things. Yeah, that there's there's some non-recoverable issues that can happen and I'm I I don't want to doing that stuff.
Tamara Laine: Yeah, well like I I just recently set up we we have a new company naming system for all of our files, right? Just because as it grows, you know, too many people are putting things in different places and you know, we had a a wiki at one point and now we we're using no like all the all the different things that go along with that. So
James Carman: Yeah. Yep.
Tamara Laine: I created I created a skill in Claude to just go through and make sure and rename and and name my new file files and all of that so that I don't have to bother with it, right? which is right, one of the real case points for what a good agent does, right? But this this skill kept trying to delete files. And as I and I was cause I was trying to get it more on auto mode, right? I was trying to get it more towards and so like I it was just like now I can't I have to back up, I have to kind of rework this thing because it just keeps trying to be like, this is a duplicate, delete it. And I'm like, it's not it's not a duplicate just because Just because someone doesn't name it right does not mean it's a duplicate.
James Carman: Right, yeah.
Tamara Laine: And, you know, and sometimes, you know, duplicates are fine if you've got them in different places for different reasons. I just there was the whole the whole thing. I was like, no, you can't. Cause like we, you know, we have a data or drive. And so like it was trying to delete things in the data room because it was a duplicate. And I was like, no, stop. You're and just wasn't getting it. So I had to I had to step in. And so I could see where like as simple as that should be, it it turned out to be really something I
James Carman: And the other thing that gives me pause with something like this. So I can go through and I and I I was texting our our producer over the weekend, Ryan, and just like, what is wrong with Opus right now? Have you noticed this? It's like it goes through like these these eras or something where the like the models are just it was just awful. Over the weekend I I I kept having to like undo code that it was writing and and like wow, wait a minute, we decided this 10 minutes ago. What are you doing? Like It was it's been awful for like two weeks. And that scares me because we could you could have exactly like you're talking about, I've got this process dialed in. It's it's identifying the duplicates the right way I won and blah, blah, blah, blah, blah. And then two weeks go by and Claude gets a lobotomy or something. And then and then it starts making really dumb decisions and deleting all my files, which I thought was working perfectly. you know, that goes back to, you know, how do how do you test these systems and all of that and evals and all of all of those fun things. So Crazy. Okay.
Tamara Laine: Yeah, it is it is funny how all of a sudden you're like, why is this not you know, even with like writing tone, right? I've got
James Carman: Mm-hmm.
Tamara Laine: I've got some of my agents really set up to my tone. I've ingested all the podcast transcripts, right? Like it knows how I speak, it knows my it it it knows my tempo, all of that. And then every once in a while it'll send out I'm like, I you know, like I've told you no dashes ever. Like
James Carman: That's the one that gets on my nerves. Yeah, I hate
Tamara Laine: Yeah, I
James Carman: those.
Tamara Laine: you know, and or you know, it goes back to something that's completely predictable and you're just like, What? What happened? And you're, you know, I've got to retrain it again. So yeah, it's weird how just out of the blue it'll do
James Carman: Yeah, and I I've got the, you know, use the Oxford comma. Like you can you can take my Oxford comma out of my cold, comma, dead, comma, and lifeless hand, right? You know what I mean? Like I I am sticking with the Oxford comma. Those sort of things. And then it'll mess them up. Like you you you tell it those the same thing. M-dashes, Oxford com like anyway. I'm with you. It does it it just for some reason it just it just forgets things and doesn't doesn't abide by your rules anymore. Just out of the blue too. mumbling next segment of our show is what we like to call. Our lightning round. And the lightning round, I always have to warn people because you know, everybody hears lightning round and they think, this is going to be fun. It's not fun. This is serious. This is where the answers matter. This is, you know, most people they'll fast forward through all the AI stuff and we're talking about governance and and you know, how do you worry about, you know, making decisions that I can lend people money and that like they're not gonna listen to that mumbo chumbo. They're tuning in for the The where the rubber meets the road, and this that is our lightning round. So the answers are correct or they're not. And I don't know what the answers are. I I don't know. We have an algorithm that we use to score this, and it takes months for it to do all of its number crunching and all of that. we'll get we'll get your score to you eventually. I just I just have to make sure everybody understands because when we say lightning round, that always sounds very kind of whimsical and everything, but it's it's not at all in
Tamara Laine: Serious.
James Carman: any way, shape, or form. It's serious is serious. And you know, this this could be life altering, you know, this sort of score. it's it's it's really it's it's really important business. So I want to just make sure that you're framing your mind around our lightning round and make sure that you're you're ready and you're mentally prepared. Are you ready? Okay. Good. Good, because I think you're gonna see what I mean. real hard hitting stuff. Do you Instagram your food?
Tamara Laine: I used to now no.
James Carman: No. Yeah, my kids are they do that stuff. They yeah, the phones gotta eat first, you know, that kind of stuff. I usually wait so long to eat that I'm I'm I'm gonna kill someone. I give me the food. Don't no, the camera can wait.
Tamara Laine: I kids now, I eat standing up.
James Carman: Yeah, yeah, exactly. In between, you know, volleyball
Tamara Laine: Mm-hmm.
James Carman: lessons and volleyball practice and vol yeah, mostly volleyball for me these days. have you ever
Tamara Laine: No No, I don't think.
James Carman: Good answer. I feel like that's a good I don't know. Like I said, I have no idea.
Tamara Laine: I I'm trying to think. I mean, maybe something from my brother or, you know, my mom's closet, but nothing serious.
James Carman: sourdough or wheat?
Tamara Laine: Sourdough all day every day.
James Carman: Hmm. Now do you make your own? Are you one of those folks that has like your
Tamara Laine: I we've been trying
James Carman: starters?
Tamara Laine: we've been trying to find a starter somewhere because it's like a pet, you gotta feed it. we're still working on
James Carman: Mm-hmm.
Tamara Laine: that. I'm I'm out in the country. We would love to have our own sourdough, but I go buy it from a local farm.
James Carman: Hmm. Yeah, that's the best way to go. Cause I'm with you. Like trying to like like it's almost like having another kid, you know, having this sourdough in your fridge. Like, I got to feed it every day or whatever. Yeah, I can't I can't do that. do you believe in favorite junk food. What are you eating when you're re reaching for some junk food?
Tamara Laine: my goodness, besides if my my junk food is is I would say wine, that's my vice. But if it's a junk food, I would say cheese dip. Cheese dip all day.
James Carman: Are you a are you what kind of cheese dip are you going for? You going like with the beer cheese kind of thing? What are you what where you going? Hmm. Yeah, it's good.
Tamara Laine: Queso, queso dip. Yeah.
James Carman: name a primate besides monkeys and apes.
Tamara Laine: isn't not a whale, isn't like a manatee one? No.
James Carman: I don't know. I d like I said, I don't know the answer to these things.
Tamara Laine: Trying to I'm trying I'm trying to think outside of the box here,
James Carman: We'll see what the algorithm says. I wish I'm like Richard Dawson here. Yeah.
Tamara Laine: All right.
James Carman: Like, or I guess the new guy. What's the what's the new guy on Family Feud? You know, survey says, right? You know, that kind of a thing. I I don't know. We don't have the board up to to to check these questions on on the fly. we can't afford that. We're a low budget operation here, so we'll we'll get there one day.
Tamara Laine: Wait, did you say private an ape? A gorilla would be a a primate, right?
James Carman: yeah, yeah. Primate, yeah, yeah, yeah.
Tamara Laine: A gorilla. I've and I've actually gone gorilla trekking, so g I'm gonna go with gorilla.
James Carman: Have you really? Like you've been up close and personal with him? Wow.
Tamara Laine: In Rwanda.
James Carman: Wow. Man, you've done some cool stuff man. We just have to like stories with Tamara. Like, we're gonna have to do a whole series on this because you've had some really cool stories that we could hear about. was it scary? I would imagine that's pretty scary. Gorillas are they're big.
Tamara Laine: Gorillas, yeah. What turns out is scarier. I I didn't realize that like 10 people a year die on that track. and sometimes it's from like stepping on hornets. There's like killer hornets' nests that you could like step on, or they're all sorts, yeah. I didn't realize that
James Carman: Wow.
Tamara Laine: when I went. Otherwise I think I would have been a little bit more cautious, but it was it was
James Carman: Mm.
Tamara Laine: really an incredible experience.
James Carman: And hippos. Hippos is the thing I'm scared of. I hear they kill a lot of people. Like they're dangerous. Yeah. Yeah.
Tamara Laine: Yeah. Yeah. I I I think it was an AI video, but I saw like a hippo chasing a safari truck.
James Carman: Yeah, yeah. And the one that like in the in the water, it may have been AI too, but it's like hopping out of the water, like leaping out, like chasing the boat on the at like, how in the world has can this huge thing do that? All right. What about this is we've been asking this question a long time for months on the show, and it's it's been an interesting journey for folks. Do you like the smell of gasoline?
Tamara Laine: No.
James Carman: No, just flat out, not ever. Just no. Okay.
Tamara Laine: No. Just no.
James Carman: Do you like the name Charlie for a girl?
Tamara Laine: I could do Charlie for a girl, yeah.
James Carman: Well, I feel like that's a good answer. Have you ever tasted soap?
Tamara Laine: I mean, yeah, I think any kid who grew up in the eighties, someone put soap in their mouth. Someone put a soap bar in your mouth at some point. That's Yeah.
James Carman: Mm-hmm. Yeah. Yeah, we all we all had to do it because Yeah. That's how you learn the boundaries. You you say that F bomb or whatever, and and your the mom's like, no, you didn't, right? Like,
Tamara Laine: Yeah.
James Carman: no, eat this soap for a little bit, and then that'll that'll clean that up for you. Did you go to sleep away summer camp as a kid? Did you? Like
Tamara Laine: Yeah, yeah.
James Carman: as was that like an every summer thing, or was it just once was it a
Tamara Laine: Yeah, it was it was a handful of summers. I went to like
James Carman: Okay.
Tamara Laine: this horseback riding camp when I was younger, which was fun.
James Carman: We sent my oldest to one and and she had never ridden a horse and she went and she rode horse and she just fell in love with it. She loved it. She was like, my gosh, I love horses. So but now she's she doesn't do anything with horses, which but that was kind of a cool cool. Are you a dark chocolate or a milk chocolate kind of person? If you're journaling, this is an interesting because you're a journalist, or a former journalist. If you're journaling, are you gonna reach for paper or are you popping up a computer, iPad, some sort of digital format for your journaling needs? Are you are you going paper? What are doing? She's got paper right in front of her. She picked it up and held it up. That's
Tamara Laine: Paper. Yeah. Paper.
James Carman: I find a lot of people do that.
Tamara Laine: Right in front of me.
James Carman: Yeah. I and I always have a there's nothing in it right now because I was, you know, just leave a page there for me to write things. But I keep a paper journal, right? handy in case I need to write things down. I'm with you.
Tamara Laine: It's really good for the the the memory brain connection.
James Carman: Yeah, that tactile. Yeah, I think, yeah, I'm with you. Okay. And finally, again, we save the kind of the most consequential things. a lot of times we'll we'll like to save that for the last question. what temperature do you set your thermostat at?
Tamara Laine: it's a fight in my house. If I had my way, it would
James Carman: Mm-hmm. It usually is.
Tamara Laine: it would be seventy one to seventy two, but I think it's usually at about sixty eight. Yeah. Yes. Yeah.
James Carman: sixty eight. The husband likes it colder. You like it warmer.
Tamara Laine: I like it more. Yep. Yep.
James Carman: I like it colder too. I do too. I and that's that's interesting because when I ask that question, a lot of folks it's it's well if my you know so and so sets it, it's this. If something if I said it, there's all that's usually the type of answer we get. and our our director, producer, Ryan is saying seventy two is perfect. Well I
Tamara Laine: Yes. So that's right. Seventy one, seventy two. That's right.
James Carman: I think the problem for us, my office is on kind of the main floor of the house. My wife's is upstairs. So we took like a one of the bedrooms or something and converted it into her office. So she's a little warmer up there. So if she'll come down and some days it's too hot, she'll set it, you know, lower, freezing me. And if I'm telling you, I I can't. I need like my my fingers get stiff when it's too cold. Yeah, first world problems, right? Okay. All right.
Tamara Laine: Yes. I will say though at night I like it a lot colder. I do tr I do try to take the Bernie
James Carman: Yeah. yeah, you could get stun.
Tamara Laine: Sanders approach of much colder at night.
James Carman: I thought you were talking about like what he was sitting in that chair, he was all bundled up in that chair. Yeah, that that that Bernie Sanders approach.
Tamara Laine: no, apparently there were stories that like, you know, for because for good sleep, you're supposed to be a little bit like bun like bundled but cold air. And like Bernie Sanders when he was on the campaign trail, like he man it was like something he mandated that his hotel rooms had to be like sixty four or something. I can't remember the exact story. I'm h maybe I'm making it up. Maybe it's fake news.
James Carman: That sounds right.
Tamara Laine: But it sounds right, that he was very much in
James Carman: Yeah. I saw something not long ago, and I'm sure it's yeah, I don't know. Forever since I was a kid, I have had to have a fan while I'm sleeping. And a lot of times I'll have it blow on me. I read something the other day that was like, that's actually bad for you to have a fan blowing on you. It's like for something involved with like your temperature regulation doesn't work right, or I don't remember, but it was like, Freaking me out. Like it can caught like it was causing these like really bad things to happen if you have a fan on your face. I'm like, what have you been doing this fifty years? Are you kidding me?
Tamara Laine: Yeah, I I feel like there's a new study. I'm really trying to, yeah. This is this is me getting off of Instagram because every time I get on Instagram, there's a a new thing that I shouldn't eat or that I've been doing or the workout that, you know, is wrong for my body or all I just I can't do it anymore. You know, and then you look at European, you know, communities and you know, they are just loving life. They're eating whatever they want. They, you know.
James Carman: Mm-hmm.
Tamara Laine: Bread all day, as a
James Carman: Mm-hmm. Yep.
Tamara Laine: staple of life. like all the and you're just like, Well, is it just our stress level? Maybe if it was if we just stopped stressing so much.
James Carman: When you hear people talk about like, yeah, I went to France and I ate everything, desserts and blah, blah, blah, blah, blah, bread every single day and escargot and all this stuff. And I lost weight. Like, what?
Tamara Laine: Yeah.
James Carman: Like, we're doing something wrong here in the States.
Tamara Laine: It's a hundred percent. That was a hard lightning round. Did I pass?
James Carman: Well, like I said, we and we'll get the scores tallied up. We'll have to run it through the system. And we there's a we got to take the transcript and there's a there's a whole embedding process. I I don't even understand all that's going on there. But they tell me it's pretty complex, is what they say. Cause it takes a while. Now I don't know if that just means the people that we have doing that act like we got a crew. We don't have a crew. We have we low budget here. They're just lazy. They don't they'll get around to it. Like I don't know what's going on there. but that's what they tell me. I whatever. but you'll get your score. Yeah.
Tamara Laine: Well I'm very competitive, so I wanna win.
James Carman: yeah. I mean, I feel like I I I like your chances, I do, of of winning the the lightning round. And we will we're we're putting together this badging program where you can actually share it on LinkedIn so you can, you know, brag about your your forward slash score from the lightning round with to others as a credentialing mechanism. Because it's important, as I said. and that'll that'll come in handy somehow. I don't know. But I'm I'm assuming it will. All right. Tell us about anything. This is in closing here. We've we've we've talked about some really interesting stuff. As I said, you are in the thick of it. You're building the kind of systems that everybody wants to try to build, like the things that are making important decisions and doing it wisely and safely with using AI. So you're you're at the forefront. I love it. So thank you for coming on for that. Anything you got coming up? product releases. Are you speaking anywhere? Do you have a a book coming out? Are you planning on starring in any movies where you'll get an Oscar? We'll have to have you come back. Or a t or a Tony. Or a Tony. W either one.
Tamara Laine: that would be delightful. Tony. no, but if anyone ever was trying to make a documentary on something cool and technology, I would be willing to to figure out how to help that. so that's an that's an open offer. because that's just a a passion. If I can if I can put together a couple of passions, I I'm always happy. so what's coming up for me? I mean, I'm the keynote speaker at a couple of conferences
James Carman: Okay.
Tamara Laine: for for credit unions. So I can give you the details on that. One coming up is in New York for the New York Credit Union Association. I'm gonna be talking about hallucinations and legal things that you should look out for. and then what?
James Carman: NYCUA. Is that is that what do they call it that? NYCUA NY Secret Union NYC I mean I don't know if that's the thing. It's just I thought that was an interesting acronym.
Tamara Laine: Yeah. I my podcast is going to be so What the Frack, my podcast that I believe you're gonna be on soon, is going to be
James Carman: I would love to do that.
Tamara Laine: is gonna be broadcasting live from the Boston Startup Week podcast stage. So
James Carman: wow. cool.
Tamara Laine: on September 15th, we're really excited about that. we're gonna have some some real fun on the the Boston stage. And yeah, and I mean our our product is launching the end of this month with our we've got several customers lined up, our first pilot customer. So we
James Carman: Nice.
Tamara Laine: are we are heads down and getting everything wrapped up to be live with them. So that would also mean officially we're live.
James Carman: Gotcha. All right. So that's that's pretty that's big. Is it is that scary?
Tamara Laine: Yeah, it's huge.
James Carman: Is this is it a scary thing or are you excited? Were you more excited than scared?
Tamara Laine: More excited than scared. but I think any any founder should always be, you know, like I n it's never good to be overconfident.
James Carman: Right. Yeah. Yeah. Try to see around the corners of what could be going wrong. Now you
Tamara Laine: Exactly.
James Carman: mentioned like you would help someone with a documentary, you know, think about do you miss it? Do you miss that life of the journalism thing? Is that is that?
Tamara Laine: yeah. I mean, yeah. I I think that there's there was just something so incredibly vibrant about every day in that lifestyle, right? Of just
James Carman: Sure.
Tamara Laine: like kind of voracious learning. and now I'm you know I'm doing the same, but I'm I'm doing it very much in in one place. And I find as a founder I do a lot more digging into to some stuff. So I I think that there's there's some times where I'm like, I would love to be out doing stuff. And I should also say I was really lucky. I gotta do so many incredible things while I was doing journalism. Right? Like you've heard a handful of them. I gotta drive
James Carman: Yeah.
Tamara Laine: a NASCAR. I went skydiving. I
James Carman: my gosh, that's so cool.
Tamara Laine: yeah, I mean I I just I did some things. I went on an there was a snowstorm that like hit the northeast and I went out with an emergency crew all night overnight to just kind of like just see how they're saving lives. And it wasn't a hard hitting story, but it was really impactful because you're out with this EMT team while it was snowing three feet and just like seeing them rescue people.
James Carman: Wow. Man. Now how do you replace that in your life? Like that's pretty that's pretty cool. That's cool stuff.
Tamara Laine: Yeah. Yeah, no. It's
James Carman: Yeah, you can't like you can't like join the HOA board or something to repla you know what I mean? Like, Sally's got her grass too high. No, let's
Tamara Laine: Hey.
James Carman: let's dig in here. Like that's the there's you can't it doesn't compare, right?
Tamara Laine: Yeah, no, I but there's a time and a place for everything, right?
James Carman: Yeah. Yeah. That's really cool. Well, again, as I said, you're you're out there, you're on the on the cutting edge of of of a lot of the things that a lot of people are talking about and you're doing it. So that we appreciate you coming on and sharing insights. how do people get go look up empower and and the what what the frack is that on Apple and all of all of the places?
Tamara Laine: Yeah, what the frack is on all the places. YouTube is where we focus a lot of our attention. But if you want to get in touch with me at LinkedIn, DM me on LinkedIn. That's the fastest way. I'm I'm a little bit behind on my responses this week, but like that's that's where you know you can chat with me. I'd love to connect. and then empower LinkedIn and our website,
James Carman: not E P O D.
Tamara Laine: no, so mpwrai.com.
James Carman: okay. No I thought it was a dot AI. That's everybody's doing the dot
Tamara Laine: Yeah. I wanted
James Carman: AI thing.
Tamara Laine: I wanted the dot AI. Someone someone grabbed that up and it was really expensive. So I just went, all right. yeah, no, someone is definitely just squatting.
James Carman: Are they just squatting on it? I hate that. That's that Ugh I hate that. Well, again, thank you so much for coming on. I really enjoyed the conversation and hope to hear hope to well hope to join you on your podcast. But you know, come back again. We can do more stories about all the cool things you've done in your life. That's that's what I want to hear about. But thank you.
Tamara Laine: Love this conversation. Thanks for having me on.
James Carman: All right, if you'd like to get in touch with us here at the forward slash, drop us a line at theforwardslash@callibrity.com See you next time.
James Carman: Yeah, that's fair. Welcome to the Forward Slash Podcast, Welcome to the Forward Slash Podcast, where we lean into the future of IT by inviting fellow thought leaders, innovators, and problem solvers to slash through its complexity. I am your host, James Carmen, and today we're talking to Tamara Lane. Tamra Lane spent years as an investigative journalist and won an Emmy digging into ethical AI. She was also part of the 2012 UN campaign to ban autonomous weapons, standing alongside Nobel laureates making the case that some decisions should never be handed to a machine. She now builds autonomous systems for a living. Tamara is the founder and CEO of Empower, the first policy bound agentic infrastructure for lending. Tamara, welcome to the show.
Tamara Laine: Thank you for having me. Quite an introduction. I should say
James Carman: Yeah.
Tamara Laine: I was I was j merely assisting the UN campaign that the Nobel laureates were doing all the hard work. I was very
James Carman: See, I was gonna ask, I thought you were in the fighter jet that was going against the robot fighter jets to like pro like you were like you were the maverick of the yeah.
Tamara Laine: When I was an investigative journalist, I did do a air to air refueling mission, which was incredible. Yes.
James Carman: That's gotta be cool. Okay. All right, now we gotta talk about this. So all right, how
Tamara Laine: Yes.
James Carman: did it happen? How how how did you find yourself in this this c this scenario where you would need to do such a thing?
Tamara Laine: it was a time I if you if everyone remembers or maybe younger people won't remember this, but back in like 2012, 13, 14, a lot of gear, weapons, were being kind of sold or auctioned or distributed to police forces.
James Carman: Mm-hmm.
Tamara Laine: yep, so you remember all of that? A lot and there was
James Carman: yeah.
Tamara Laine: It was very controversial and so I covered that a lot. And part of covering that I got into a lot of conversations with the US military and was invited to do a few things with them that were really incredible to see firsthand, just really the skills, the training, everything that goes into some of the US Air Force missions as well as the military missions. it it is it is quite incredible. So kind of putting aside the the in the US controversy of U US police forces having military weaponry, white right?
James Carman: Mm-hmm.
Tamara Laine: Looking at how the military operates and and that was really incredible firsthand to be up there in the air. I also did some MRAP training and it was quite incredible. Yeah.
James Carman: That's awesome. That's so cool. All right. So you've you've had a really, really interesting background. I'm so happy to have you on the show because this is you have such a such a cool background. So you did investigative journalism for a while. You're digging into all sorts of stuff. And you do this story on ethical AI at some point. And that was that the trigger for you to kind of convert to to building these autonomous systems yourself, or did this happen sometime later? How did this journey happen?
Tamara Laine: The journey, so one of the media outlets that I was a part of helping launch had to do with impact investing. And what I was doing in particular was talking to founders and really looking at the startup scene. And that's where I started to fall in love with building technology, building startups, seeing how you could change people's lives. I I was working kind of how do I help people from the journalism side to how do you you as a company corporation change people's lives. And so I really became interested in that. And so when I left journalism, I I wanted to go into that zero to one phase. I I thought that it was incredible and had so much potential. but my life really collided with AI when I did this story on activist investors fighting for a code of ethics in Amazon and Northrop Grumman's technology AI and their AI development.
James Carman: Mm-hmm.
Tamara Laine: What I was doing was I was talking with ethicists and and technology drivers and and you know investors. I was talking to the whole spectrum of people who are involved in this one particular area. And I I found it just absolutely incredible. And that's where I really started to dig in and kind of continue down that path of AI and how it was going to interact with our with our world.
James Carman: That's that's just a fascinating story. The most people get the get the bug, like the the technology bug kind of early, but you you kinda had a different journey. I love it. That's awesome. and this this was, as your bio said, an Emmy winning invest investigative
Tamara Laine: Yeah.
James Carman: journalism. What what how I don't know how does what kind of Emmy is that? Like what what is the the official category?
Tamara Laine: It's a
James Carman: How yeah.
Tamara Laine: it's an Emmy. It was my goodness, I'm gonna forget the category. It was it was like short form documentary. I have to look back at it. I'm forgetting the the category that it was in, but it was it was a longer it was a longer form story. It wasn't a full documentary. So it was about
James Carman: Gotcha.
Tamara Laine: seven minutes. so it was like it was a special expose, but I just off the top of my head I'm now forgetting the actual category. If I I have it here. Do you want me to grab it?
James Carman: Yes. Well we want to know about this. This is cool stuff, yeah.
Tamara Laine: Okay. Sorry. Here it is. eyes on you, keeping
James Carman: Gosh that's so cool.
Tamara Laine: recognition companies in short short form content. So technology short form content up to ten minutes. That was it.
James Carman: All right, so this is so cool. She's got this I you can't see it, but she's got the statue. She's holding it in her hand. Like that. This is I
Tamara Laine: Yeah.
James Carman: I don't know if I know a lot of people. I think we've had guests that we've had one other guest that had an Emmy. I think it was a daytime Emmy for something related to public television or something like that. And then we've also had someone who had a Grammy. one of our folks, one of our guests had a Grammy. So we're we're still looking for that Oscar and the Tony. We gotta get we're we're going for the egot, but but That's that's really cool. now how did and I gotta I gotta geek out on this. So did you like go to the Emmys did like to receive the award or was this one of those like well it's short from life mail it take? How did this work?
Tamara Laine: No, that well, this was such a bummer. it was the it was COVID. So I sat on the couch, dressed up with my family, and my daughter thinks that she won the award. She was very proud of herself. Yeah. Such a young age, so impressive.
James Carman: well that's good. Good for her. At such a young age, even. That's impressive. Yeah. Yeah. So they did they like film you like when the when the announcement happened and everything? Like did you did you have like your phone?
Tamara Laine: No, it was just so sad. No. yeah, but that's okay. I still you know, the accomplishment is what it is. Not the not
James Carman: Yeah.
Tamara Laine: not the evening. And I went out for drinks with my producers later and it was a a good time.
James Carman: That's really cool. That that's that is such a cool story. All right. So the the ban killer robots is the campaign, right? That was the
Tamara Laine: Yeah, the campaign to ban climb.
James Carman: Okay, tell me about that. Like what what what all what was that all about? I want to hear more about this.
Tamara Laine: Yeah, well that was again so very early in my career, and and what it was was it was several Nobel Peace laureates and and people, you know, basically academics that had been studying this and as well as people in the military who kind of came together, or former former military came together and saw that this movement that technology was becoming incredibly smart, right? And what they wanted to do was get the government to put in regulate like controls, control levers around autonomous weaponry, right?
James Carman: Mm-hmm.
Tamara Laine: And so they were going all over talking at you know different government events. when I when I was volunteering in this and and working with part of the team, we went to the UN. And we sat in a forum on this and discussed, you know, its it its significance. And this was back in 2012. And so kind of where I find irony is back then when we were talking about it, people were like, this is not gonna be a huge deal, right? Like, this
James Carman: Mm-hmm. Mm-hmm.
Tamara Laine: isn't this isn't my problem today, right? They ever was just like, you know, like w no one's gonna let autonomous weapons just roam around. And they were already, right? Like there was autonomous, guns actually on like border control. you know, our our drones of that time could be autonomous, but we just didn't let them. so it was really a a significant conversation to to happen. but I just I remember sitting in these rooms and we were having these discussions and just kind of people being like, like that's not gonna happen. And now if we're, you know, we're 10, 12 years, 14 years later, it's not like Everybody, if you pull people in the US, they're like, yeah, right? Like, yeah, we've seen we've seen the robots walking around. Like I'm in the Boston
James Carman: Mm-hmm.
Tamara Laine: area. Like at our children's museum, the Boston Dynamics,
James Carman: yeah.
Tamara Laine: you know, robot that can walk over any terrain across the world. It's just the it's like moving around in the children's museum, right? Like it's there. So like we're watching what's what's that championship in China right now of the
James Carman: Saw that, yeah.
Tamara Laine: the the robots, I for what's the name of it there? The two legged yeah, the two it's like
James Carman: It was like a Robot Olympics kind of thing, wasn't it? Yeah.
Tamara Laine: two legged robots and just to see and like they can run faster than humans. It's in it's wild. It is wild how far we like the technological technological advancements in the last twelve years, which is just it boggles my mind. But yeah, so then and I remember again I was super young back in this when I was working on this campaign. I guess I I wasn't super young, I was super young in my in my understanding of this and my work in it, right? I would say I was pretty, pretty junior. and I was sitting at the UN in the round in in the circle. And because they knew I was a journalist, the team asked me to ask the the ambassadors at the table to like raise my hand and be the one who brought this up. And so I I
James Carman: Nice.
Tamara Laine: brought it up to like the ambassador of Egypt. like raised my hand and and started the conversation on, hey, what are we doing about autonomous weapons and, you know, regulate I forget the question, but I I do remember the the s the situation because I was just like, you know, it was I was also very early in my journalism career. So I had sat across heads of states before, but it's, you know, when you start asking those questions young in your career, you're You get a lot better at it when you get when when you get your your C Lane.
James Carman: How was that received with with like, you know, just this this person, a journalist asking these heads of how was that received? Were they like who are you to ask that? Or were they were they like just answer?
Tamara Laine: Well, no, because you're sitting around the table, so they just answer it, right? Like it's not
James Carman: Okay.
Tamara Laine: yeah, they didn't think I don't think there were people thought much of it other than just like you've got to seat at the table, so you you've kind of earned the right to ask questions. But yeah, but inside
James Carman: Nice.
Tamara Laine: I was like, Whoa, don't screw this up.
James Carman: Yeah. my God. Wow. Man. All right. We we could just talk all day about this stuff. But I but you're here to talk about AI and and what you're doing now. And what you're doing is what everybody's talking about. Trying to build safe, guardrailed,
Tamara Laine: trusted.
James Carman: contained, trusted, all of the things, right? give us those warm, fuzzy feelings about our AI agents. That's what you're doing. That's what you are doing with empower. That's that's what the kind of is that the mission of the of the company is. you know, kind of building safe and
Tamara Laine: Yeah, I mean, well, we're building in highly regulated industries, right? So we're we
James Carman: Mm-hmm.
Tamara Laine: are the policy engine for for orchestration and underwriting right now is our is our niche and know that we'll expand. But what we are what we are doing is we are building highly guarded, you know, regulated agents that are in an orchestration a layer that can do the work. for right for sorry lenders. I was gonna say regulators. for lenders. And that's real, really challenging because I don't trust AI. Right.
James Carman: I don't think anybody does right now. That's that's the problem. Yeah.
Tamara Laine: So yeah, I've got a really incredible team and we're creating a product so that we can really you know the the underlying goal is improve access to capital for for everyone, right? So improve the lending process for everyone. And so our agents, what we're doing is we're building the policy engine and the infrastructure so that agents can do all the work and orchestration for a highly complicated, highly regulated task. And so in order to do that, you do. You have to, you have to build, you know, it has to be transparent, auditable, you have to abide by all these rules. The guardrails have to be very secure, right? Like it's it It's not for the faint of heart.
James Carman: Yeah. Yeah. I wanna it's interesting because a lot of people are are talking about this stuff in a in a very kind of hand wavy motion right now. There's like, well, we gotta put guardrails on these things. Yeah. So
Tamara Laine: Airbase creating agents. Yeah.
James Carman: we're gonna create agents. We we got we know we gotta put guardrails on this stuff. What yeah, we we're gonna make it safe. It's very surface level conversation, but you're getting down into the meat of it. So it'd be interesting to hear your perspective on kind of what are some of the the key aspects of of of this workflow. Like, you know, key components of of being able to make a good decision that is auditable, that is traceable, that you can you can back up and you can say, Yeah, this is why you know, all of all of the things, right? So may me can we break it down maybe into some componentry here? Just some low level like you you gotta get this part right, this part right. Can we do that?
Tamara Laine: my goodness. Let me let me think of how to how to answer that. because I we're in the we're in the weeds right now, right? So like
James Carman: Yeah, yeah, yeah.
Tamara Laine: so ask me that question again just so that I make sure I'm I'm hitting it correctly.
James Carman: Well, what are what are some of the well first let's start with some of the failure modes, maybe. Like, so okay, an agent in your domain is is trying to make a decision based on is is the idea basically someone can or cannot borrow money. Is that the idea?
Tamara Laine: Yes.
James Carman: In a very simplistic Yeah. Is that what you're saying? What w that's the domain where
Tamara Laine: Yeah.
James Carman: where you work?
Tamara Laine: Yes, exactly. Can or can so our agents are helping parse through the data to give to an underwriter to make a very fast decision on whether or not someone ki in within the institution's policy qualifies for a loan. So
James Carman: Okay.
Tamara Laine: we've got a lot of working parts on that.
James Carman: Yeah. I know there's and there's there's a lot of regulation. I know is it isn't it Colorado has got some stuff that's activating. Was it January? They pushed it back to January of twenty seven. Is that right when they're activating? Yeah.
Tamara Laine: Yeah. And and you see the same thing in the EU. Like we're actually really following
James Carman: Mm-hmm.
Tamara Laine: the EU stuff because that usually is a little bit more progressive than any of the state stuff.
James Carman: Yep, yep.
Tamara Laine: and stuff is coming out of the Treasury Department as well. But yeah, I lending and underwriting is considering the most high risk task
James Carman: Mm-hmm.
Tamara Laine: for AI. And so what you have to parse through with that is to make sure that no and and actually I just firmly believe AI should never make the decision. So what AI needs to be doing is the the grunt work, right? It needs to be gathering information, going back and forth, doing a lot of the communication with the borrower, like missing documents, standardizing documents, right? Like when you're getting documents from multiple dis different sources, what is the source of truth? Where, how do you how do you solve conflicts and information and with what with what trust value?
James Carman: Gotcha.
Tamara Laine: And working in a highly regulated, a lot of people, I think this is my my controversial take, but a lot of people do not are pushing for less regulation in AI. I find that having regulations, firm regulations, right? Standards that you have to operate off of give us a lot more guardrails to keep things safe, right? And
James Carman: Mm-hmm. Yeah.
Tamara Laine: that way, that way we have to operate in a safe way. We have, you know, like there isn't wiggle room for you know, bias. We we can't. We just can't, right? We can't shift something that is biased, where a lot of other industries can, right? even in security, you can do that, right? Like it's just it's I think it's really good. And having something that's firm that we know we have to operate is way better than having some ambiguous rules. That we don't know which way the wind is gonna blow and how it's gonna affect us, right? And I I say that's the killer of a startup. if a startup has ambiguous regulations that they're trying to figure out, then funding starts to dry up. You know, their product can be very vulnerable if the wind blows the wrong direction. So, like I enjoy working in a highly regulated area.
James Carman: Okay. Yeah, I think I I I love this analogy. My wife and I went to a parenting seminar. It's I'll circle it back to make it relevant, but it's it was a parenting seminar and the guy told a story about a study they did of of these kids. And what they did was they would they would have these children go out on a playground and there was no fence around the playground. And when there was no fence around the playground, the kids would kind of they would hover near the building of the school and they wouldn't they wouldn't kind of go out and play on the the different pieces of equipment or whatever within the playground and they kind of hover near the school and they wouldn't, you know, venture out. Then they put a fence around the playground and the kids were flourishing and they were playing around and they were jumping all over, going on the monkey bars and all that. So that it kind of lends itself to that idea of what you're talking about. When we have those guardrails and those boundaries, it does allow us to know here's our playground. Here's where we can play and we can play freely and we know we're safe. I agree with you. I think the big thing that people are arguing about right now is Who makes up the rules? Because you as you said, like you don't want it to be just because an election happened or whatever, you know, the the the the winds change, somebody gets in there and they want to, you know, rip the rug out of it from everybody. Like we we don't want that. We want we want a nice stable environment with good rules that that we can all play with with the same rules. I think that's the the big argument right now. But yeah. yeah. Cause
Tamara Laine: Exactly.
James Carman: that is the scary thing, and that's those are the things that have been happening to people is. They get rug pulled with, you know, a regulatory rug pull where they they have a business model and then someone just passes a law and wow, my business is now gone. I there's nothing I can do about it. That's kind of the scary thing. we're we're talking about borrowing money. I know I know this this is absolutely one of the things. As you said, it's you you don't want it to make the decision. And it from a regulatory standpoint, I do think the wins are already there of of saying, like, you're not, you're not gonna make decisions about, you know. Does this person deserve to get healthcare? Are they are they gonna get a loan for a home or w get a job? There's they've already carved out like these are some of these key life-changing decisions for humans that we're just not gonna allow AI to do. so that's good. Yeah.
Tamara Laine: Yeah. Yes.
James Carman: I and I love what the way you kind of phrase that. Tell me more about this, the grunt work thing, because I think that's that's a that's a huge pattern in this in this agentic world is. Go do all the work and gather up the body of work and put it in front of a human being and let them make the ultimate decision. So say so say more about that. How how does that work?
Tamara Laine: Yeah, so the idea is to take take away the busy work, right? The the back and forth. I mean, so for us, we're pulling data sources from hundreds of different we're pulling data from hundreds of different data sources, right? and for for you know, for lenders and kind of caveat like, you know, every lender is different, has different capacity and and different IT teams and all of that, right? but kind of inevitably that's the part that slows down every process. So if you look at
James Carman: Mm-hmm. Mm-hmm.
Tamara Laine: mortgages, like mortgages can take between 30 and 45 days. And it's not because it actually takes that long. It's because things are passing through so many different systems that it just drags. It's called like exception drag. And
James Carman: Mm-hmm. Absolutely.
Tamara Laine: so Asians can just do that more efficiently, right? They can pull the information, they can parse through it, they can find out what's missing, they can, they can get in touch with a borrower. They can get the information from the borrower. They can file things in the right spots. Like agents can just do this in a more efficient way with the right rules and bounce put on them. And so that's what we're doing. I mean, we I I got up and I gave a keynote a while ago and I said, you know, AI should be your your trusted junior analyst. And that's how you look at it.
James Carman: Yeah. I like it.
Tamara Laine: Because we don't trust AI.
James Carman: Yeah. I yeah, I don't think it's earned it yet. So I mean it's it's it's it's true. And I don't know that it ever should for some key decisions as as we're we're carving
Tamara Laine: Yeah.
James Carman: out. I don't I don't think it's ever should should earn to it's the right to make those decisions that are truly that impactful on a human being's life.
Tamara Laine: It really I mean, like I I said that twice and joked around. I think that it is such a revolutionary technology and I think that it has really and I ha I I do love the technology, but love and trust are two different things.
James Carman: Yeah, that's fair. I think that's fair. Well, what about? So in this process, I guess the one failure mode that I can think of off just off the top of my head, and I'm sure there are others, is the agent goes through gathering information, the the body of work and says, Hey, underwriter, here's the the case file for this thing we're doing, right? say yes or no to this, and you you make the the ultimate decision. Well, what if it doesn't gather complete data? That would be kind of the one failure mode I can think of, is that it didn't give you all the information that it should have or could have for you to make a more informed decision as an underrated.
Tamara Laine: Yeah, so you have to put in all sorts of evaluators throughout the whole process, right? Like,
James Carman: Mm. Okay.
Tamara Laine: like is is this trusted on a scale you know, of one to one hundred, what value do we give this information coming through so that when it does get to the final underwriting package, you have not only like the underwriting package package, but the trust score on multiple different sources of what you're evaluating
James Carman: Mm-hmm.
Tamara Laine: against. So that you know, like again, if you're an underwriter and you're getting something that like, okay, you say it's ninety percent, you can go, all right, well, I've been using this for a while, ninety percent is usually pretty good, but let's just say it's like eighty percent, and you're like, well, I'm gonna go double check that. So that you can
James Carman: Mm. Okay.
Tamara Laine: get really, you know, a better picture and and use your own judgment of whether or not you need to go back. And is it an area that is highly important to the organization or is it not, right? Like you can you can make your own judgment on that. but it's also really important. I you asked a question earlier, it's something that's important in agentic workflows, especially with with decision and with highly sensitive information is making sure your agents don't have the keys to everything.
James Carman: Mm-hmm. Yeah.
Tamara Laine: So how you can trust things more is if agents, you know, like every agent has its own little domain and then passes information
James Carman: Okay.
Tamara Laine: along and no one, no one no one agent can operate the whole thing.
James Carman: Okay. Yeah, I've seen a lot of patterns like that where they're these multiagentic systems where, you know, and people are setting them up, like, I run my company with, you know, all these little QT named agents and all that. Yeah. Yeah.
Tamara Laine: You see it on Instagram. This is about what was it? This is my Jarvis.
James Carman: Wake up, Jarvis, Daddy's home, you know, whatever, right? Like the
Tamara Laine: Yeah.
James Carman: the I I've seen that a gazillion times. but I I do think there there is a pattern. And I think there's there there there are studies. I don't have the data in front of me, but there are studies about, you know When you when you do have these kind of almost adversarial agents working together so that they own a domain and they have their own goals that that they want to do and they kind of balance, it's like a checks and balances kind of system. And as you said, one singular agent doesn't have all of the tools available to it that it can just go out and do everything. You kind of have to ask them to ask others and they're they're guardians of their own domain. I think that's that's yeah, yeah. Yeah.
Tamara Laine: 'Cause that's where the hallucinations really start to happen, right? When it when it's too much information for an agent and they're like, you're yeah, I mean, I'm sure everyone anyone who's used ChatGPT or CLOD or any of the two has been like, Are you sure about that? And they're like, you you're right. You were good to double check that.
James Carman: You're right to ask that.
Tamara Laine: Yeah. Your your instincts were spot on.
James Carman: And it's it's both right and wrong, but it's wrong for the right reason. Like, like what?
Tamara Laine: Yeah.
James Carman: What did you just say to me? Like, and that's the reason that matters. Like, don't patronize me, Claude.
Tamara Laine: Yeah.
James Carman: Okay. You know, I got stuff to do. I I literally was doing this just this morning, so it's hitting home for me. All right.
Tamara Laine: Yeah.
James Carman: So we're we're we're using multiple agents, and I think that one of the keys, as you're saying, is is that these agents maintain their own context so they don't bring If you if you use the same model, let's say it's Opus four or Opus five in in Anthropic, and you give it all the exact same context every single time. If everybody's just throwing the exact same context and the exact same data at it, it's going to make roughly the same decision. So that the important part there is when you partition these agents out, they manage their own context. They don't bring everything, all of the same information to bear to the to the LLM to say, help me make a decision. So That that's key because it it can water it down. When you when you pollute one call to an LLM with all of this information, it's gonna forget half of it by the time it gets around to generating stuff to talk to you. so yeah, I I I okay. So that's an interesting concept. Multiple agents, different contexts, controlling that. I love it. I wanted to come back to the word, the T-word that you mentioned, the trust and and and the data that you are bringing in, because that's that's a key. That's a key failure mode that people are seeing now. And and and I think that's going to be like the the software engineer of the future is around like understanding, okay, this data is coming in, but that was generated by an LLM. So I don't know that my trust score doesn't go high on that thing. But you know, this comes from what the human said. The human being actually said this. So I trust that more. Or this can't comes you know from some factual database. I can trust that more. But this was. This was coming in from an an L L generating information. So it's it's that prompt injection idea. Like I I I don't wanna allow for someone to creatively create a prompt that generates information that now I just put a trust score on that just muddies everything up. Is that are you seeing that? Is that are those some of the problems you're working around as well?
Tamara Laine: Well, our data, so the data that we're bringing in is maybe different than other data sources that other agents might be working off of.
James Carman: Mm-hmm.
Tamara Laine: because like we're we're bringing in pretty we're not bringing in data from agents yet, right? So I know Yeah, I yeah,
James Carman: Okay. They're not generating and they're not giving their your commentary. That's not their job.
Tamara Laine: I know the agent to agent discussion and I actually have been talking in in some of my startup groups, there's a really interesting thing that's happening, which is an agent to agent bank for agent to agent commerce, right? Which is
James Carman: Mm-hmm.
Tamara Laine: I like think about the things that we're now talking about, which are like, what? That would have been a conversation three years ago. but the Yeah, what a time to be like, but you know, it's a
James Carman: What a time to be alive, huh?
Tamara Laine: real I can't remember who I was reading was talking about this, but like just for general search now, right? Like if you were If you're someone who uses any of the the LLMs to research for like writing, right? or for commerce, if you are researching for like, I want to use a new deodorant.
James Carman: Mm-hmm.
Tamara Laine: what's what's really interesting is I forget the exact number, but it's it's kind of astounding. It's like 60% of the content that LLMs are scrolling, or it's like 60 or 70%. Is is content made by LLMs now because the actual sources where you would want to get your information from, like consumer reports, right? Does not allow scrolling on their page. So right. So they they're blocking. So like just like the New York Times or any of these other sources are blocking the scrolling, the legitimate sources that you really want to get your information are blocking the LLMs from scrolling and get and
James Carman: Mm-hmm.
Tamara Laine: retrieving information. So what you're actually doing now when you're searching is you're searching and getting information from a site that probably was created by someone using an LLM, meaning how do you trust that? so that's a really interesting, you know, thing
James Carman: Mm-hmm.
Tamara Laine: to think about that like it you kind of want to go back to Google.
James Carman: Do you remember? Yeah, I'm I'm giggling here because it just I don't know. Again, I I'm a movie guy and this reminds me. Do you remember you remember the movie? I think it was called Multiplicity with Michael Keaton, where he was like, he was just so busy, he's my gosh, if I had if I just had a clone of myself, well, I'll be darn, we can clone you. And they kept cloning, and then the clone would say, I'm too busy, I'm gonna clone myself. And by the like three generations deep of clones, the guy was like just a buffoon, right? He's like shaving his tongue, you know. I mean, it was that for me was silly stuff. But that's that's kind of what we're getting to. If if these L LMs are generating information that's gonna train LM so and it's going farther and farther away from ground truth, human knowledge, where's that gonna get
Tamara Laine: Yeah.
James Carman: us? Is it gonna get worse and worse? I d I don't know.
Tamara Laine: Yeah, it it really I mean that's why like when I do a lot of my writing, even though I do use several different LLNs to like, once I compile my thoughts and like do my rough draft to polish it almost like a copywriter, I do
James Carman: Mm-hmm.
Tamara Laine: a lot of the searching on my own from from like the websites that I know and trust because yeah, otherwise you're coming back with some blog from a law firm that created it using LL.
James Carman: Mm-hmm. Yeah.
Tamara Laine: Yeah.
James Carman: Yeah. It's it is so interesting that i even even when you do set up, because I have a bit of a a workflow that I work through with using agents to help me with writing and they they do research and adversarial work against themselves. So there'll there'll be a skeptic, right? But even through all that, it's still some garbage gets through every night. And you're right, you have to kinda you have to go ahead and check those things. But at least it does it does still streamline things. But that but you're right. We you do wanna go verify some of those assertions and and facts and quotes and stuff like that that it's asserting for sure.
Tamara Laine: Yeah. Well, and this goes, this is my advice to to younger people who are entering the workforce or who are, you know, really trying to get ahead. If you are not like if you're not double checking your sources and doing that work and doing that diligence, it it's so easy to spot. and you have to train yourself. Like again, I think we've we've talked we touched on this. AI is great for experts, right? It's really
James Carman: Mm. Mm-hmm.
Tamara Laine: good if you're an expert in your domain because
James Carman: Yes.
Tamara Laine: you are able to spot and to see through and use discernment on what's what is accurate, what is not, what doesn't feel right. Let me double check this. But if you're just going through several different prompts and then handing it over to someone else to look at, yeah, usually you can see through that right away.
James Carman: Yeah, and I think it's it has to do with the the confidence that it generates things with. So if you're not if you don't have that level of discernment and familiarity with the domain, the the LLM it's not like, Hey, I think this is true, but you know, maybe you might want to check. That's not how LLMs work. They're like, This is the facts and it's completely bogus, right? Yeah, it but it's gonna tell you confidently that that it, you know, a fact.
Tamara Laine: Yeah. I was doing some legal work, trying to I I ended up the the end of the story is I ended up just getting a lawyer, but I was using all the different tools, kind of going back and forth to do some some legal contracts. And I I said, Claude gives me anxiety doing that because it would come back a critical alert. Stop before moving on. Don't do anything until you do X, Y, and Z. And I was like, God, God, God, right? And then I would take it over to The other system and it would be like, no, according to this law, you're fine. but this is now an issue. And I was just like, I was going back and forth. I was like, this is not my domain. This is not. And I finally was just like, nope, this is why you do hire some people. This is this is why
James Carman: Yeah.
Tamara Laine: enterprise software will still be alive, right? Like this, all these things. It is because there are There are experts and there are people who do things really well and they do it well for a reason because they spend their time dedicated to it and let them do their good work.
James Carman: Absolutely. Absolutely. Yeah, I th the law thing is interesting. And you're and you're right, like the personality of the different models, you know, just based on their training is it they will give you interesting answers and then they'll disagree with one another based on, you know, just kind of their principled level thinking and how they're how they're trained to think.
Tamara Laine: Yeah, but Claude comes up and it's like red or green. Like it really does.
James Carman: Mm-hmm.
Tamara Laine: It's kinda it gives it gave me anxiety. I just like, I can't do this anymore, Claude Stop. You freaked me out.
James Carman: Now it you you said something that was interesting. So so so the whole idea of like checking your sources and all that, that probably comes natural to you. And you feel like that that journalism background makes you more adept at working with these these LLMs because you you have to do that as a journalist. I mean, you're not gonna publish a story if you you're not checking your facts. You can, you know, I there's There's ethical guidelines there. I mean, there's a you know, you you need to do that. Do you think that background is helping you in this this world of AI?
Tamara Laine: I think that that background helps me with a lot of the stuff that I'm doing. Yeah. Yeah, definitely a lot.
James Carman: Mm. True. Yeah. Yeah. It's interesting. We'll have to maybe have everybody, if you're gonna be a software engineer in the modern day, take take to a journalism class, you know, just there's there's a lot.
Tamara Laine: I honestly I think journalism journalism and law are two things that really translate well because it's you know how it's how do you think, how do you test, how do you, you know, determine validity, like all those things. And I think I think
James Carman: Mm.
Tamara Laine: those two really do a good job in creating a mindset to to work and deal with AI.
James Carman: I think it it's it comes down to we talked about this, I think, on the podcast in the past, but it's it's really about it's about critical thinking. And in in those jobs, that's one thing they they teach you. You have to be able to be a critical thinker to to do those jobs well as a lawyer or journalist and economist, you know, those sort of things. Okay. All right. So interesting, interesting stuff. And and you're right in the thick of it. It's it's it's again, what a time to be alive. Figuring out how we can trust these robots to make decisions for us or not, and how to put guardrails around them and how do you partition them correctly. I mean, we could we could sit here all day and talk about this. But we have to move on to our next segment of the show. Our next segment is what we call ship it or skip it. And this is our our take on hot or not. The idea is we bring up a topic, we talk about it, and we think, yeah, yeah, I like that. Let's ship that, or no, I don't like it, skip it. that's the gist of it. One that I think's interesting, What do you think about this notion of watermarking AI assisted work?
Tamara Laine: Skip it. am I allowed to elaborate at all? okay.
James Carman: Okay. No, absolutely. Yeah, yeah, yeah. The lightning rounds where we gotta get like terse. You know, we because they're act they're yeah, yeah, yeah. You can you can elaborate.
Tamara Laine: okay. Yeah. No, everybody is using AI to help them write. And beyond that, like what what is watermarked, right? So like if I if I do most of it and then AI comes and cleans it up, right? And it still has that AI hint to it. So that's watermarked. I think this is I think it's pretty lame. I think, you know, I get content on social media that is like deep fakes, right? I I get
James Carman: Mm-hmm.
Tamara Laine: wanting to identify some of that stuff, but from now, I mean, the genie is out of the bottle. People are using it to write, people are using it for websites, for graphics, for everything. And so I I don't think marking it is going to do anything other than like It was who I forget who it was, but just wrote the hedge fund guy, very very respected. I'm forgetting his name off to my mind. It's been there's a huge backlash on it. He wrote a wall a Wall Street Journal article, or Wall Street Journal op ed. And
James Carman: Mm-hmm.
Tamara Laine: and someone put it through the AI jet the AI test. And said it was, you 100% AI generated. And like the and the guy came back and said, Yeah, he goes, I do a lot of things in my day. And writing an op-ed is not one that's gonna take a lot of my, you know, my time. It's all my thoughts. I approved it. It is everything that I think. It, you know, it's my research. It's but I didn't sit down and handcraft every single word, right? and and that's what I think is important, right? Like, is it your is it your thoughts? Is it your your thinking? And like the Wall Street Journal was like, We've known this guy for, you know, decades. This is this is his thoughts. It doesn't matter if AI helped compile it and put it together. No.
James Carman: With you. I think early on when the the whole AI thing, and my kids are still like this. Like, gosh, dad, you use that AI stuff. That's garbage. You know, it's because the schools have scared them away from AI for some, because they don't want them cheating, right? So it's kind of a cop out. Anyway, early on, there was this mentality of like, AI slop, and you know, everything generated with AI is just garbage automatically. I thought we had gotten past that, but this watermarking thing, I think, is going to be a setback for us. And again, I not that I think everybody should be just throwing around, you know, just say, hey, write me an article and just start publishing things. No, as you said, do some research, have some ideas, develop a thought and and come up with something novel that's adding value to the world. If AI helps you write the article, okay, whatever, right? And you know, at least maybe it'll be better done than what you would have done by yourself, but you're you're able to express yourself. I I think it's a setback and I I I don't like it. I'm with I I'm trying to figure out like what is what is the real problem they're trying to solve. You said meant you know, kind of the social media, some of that deep fake stuff. I mean, that that stuff is like I it's some of it's it's just you just scroll past it these days. And it it's but it's a there's a lot out there. So I I don't know what problem they're really trying to solve with that. And I'm not a huge fan. So we'll we'll see. I don't I don't know.
Tamara Laine: Yeah, I mean, would I like less noise? Absolutely.
James Carman: Mm-hmm.
Tamara Laine: do I believe that like kids should learn how to write? I I I even now try and find time to do writing myself.
James Carman: Mm-hmm.
Tamara Laine: because it's a skill you don't want to lose, right? Like it is something that helps with your critical thinking, it helps with your brain development, it helps with all of that. Like you need to do it. you don't want to totally outsource it. We have to teach our kids how to do it well. You can and it goes back to like experts use AI well. If you're not a good writer and you use AI to write, it's going to be bad writing. If
James Carman: Mm-hmm. Yeah.
Tamara Laine: you're a good writer and you use AI to help perfect, it's gonna be really good. so yeah, so I think that there are things that we need to think about outside, like there's a more important conversation to be had instead of watermarking this stuff to like really instilling in our in our kids and and University is the importance of critical thinking and writing.
James Carman: Yeah, I I I'm trying to remember. I I did some research for a talk I gave not long ago. And this was this there was a study, and I'll have to probably should prepare better for these episodes. but they they they
Tamara Laine: I well.
James Carman: pointed out if they pointed out exactly what you're saying, that they gave they had writers use AI and there was a there was a benefit, and there were people who don't write, and there's so there's like three levels of people and the the findings were interesting. It was the the domain experts using it. about subject. They're not writers, but they're domain experts. It helped them the most because they had the discernment and all that stuff. A professional writer, the quality of work that the AI gave a professional writer, someone who knows how to write really well, is I'm recalling this off the top man. My my recollection was that that was kind of it was it was not great, right? But then the people who weren't domain experts and are not writers, mm. wasn't good at all. Right. So I think that but the biggest bump was the domain experts being able to exp help them to write better stuff. That was where the biggest bump was.
Tamara Laine: makes sense. I mean it's kind of the same how I feel with coding, right? Like
James Carman: Mm-hmm. Yeah.
Tamara Laine: there's vibe coding now can make a poor to mediocre coder 80% better, right?
James Carman: Mm-hmm.
Tamara Laine: Which like covers up a lot of a lot of non-domain expertise. but then
James Carman: Mm-hmm.
Tamara Laine: if you give it to someone who is a coder and is really good at doing it, you're gonna get something that's excellent.
James Carman: Yeah. Yeah. I I mean that was one of my early findings when starting to use this stuff is, you know, it helps me tremendously because I've been doing this a long time with you. Again, I think it you I love the word you use, the discernment. I I think that's a that's a key phrase and that's that's but that's earned. That's that's experiential. And the problem we're gonna have is If you're not doing the stuff that I did for 30 years, you're not gonna get that dismant. So it's kinda hard to build that dis that's gonna be the the the the thing we need to solve for is how do we get people the experience to get the discernment to be able to do that in the in the the future when they're not having to do the the hard thing. So that's that's gonna be the million dollar question or billion dollar problem, maybe trillion. We're we're talking crazy numbers now with AI. So it's probably trillion dollar question. All right. So the next ship just recently, Claude came out with this auto approval mode. And I guess the gist of it was what they found was, or anthropic, I should say, they they found that most of the time when they presented the prompt for the human in the loop to say, Hey, I want to go do this thing. Can I do that or not? The human would just rubber stamp it most of the time. It was just boom, boom, boom. Just go ahead and do it. Don't quit bugging me. So they've turned on this this auto mode where the they have an agent itself or some form of of yeah automated process that that makes the decision is Claude allowed to go do this thing that it wants to go do? what are your thoughts on that?
Tamara Laine: I guess the the question is what is the task, right? Like I
James Carman: Mm-hmm.
Tamara Laine: I think that it's a much broader question that probably deserves a lot more review before just turned
James Carman: Mm-hmm.
Tamara Laine: on because the answer is gr is completely different with different tasks, right? Is it a so I mean, I was gonna say is it a social media post, but I would still wanna be in I wouldn't want any anything automatic with anything that's public facing. Right. is it is it internal, is it external? Is it financial? Is it social media? Like why so many questions for me to just have a really generalized, yeah, bump it, you know, go ahead and automate this and and run with it. so I don't think that's an easy one for me to say ship it or skip it, because it's kinda like, I guess I guess I would say skip it because there's too many questions in my head.
James Carman: Too much. Yeah.
Tamara Laine: Yeah.
James Carman: It's interesting to me. I I so I would be hypocritical if I was to say completely skip it because for a long time now, and again, I'm doing, you know, research type work and and that sort of stuff when I'm doing coding these days. I'm not doing, I don't, I don't do a lot of client work these days, but most of my stuff is in R D. anyway, when I fire up Claude, I'm firing it up in YOLO mode, right? Like I don't, I don't want you asking me a bunch of questions. So that's what I've been doing for a long time. So I think I haven't really had any problems. Then I don't know. Maybe, maybe it's offloading all my information to some Chinese company or something. I maybe that's what's going on. I just don't know. but but it I haven't had any issues with it. So I think there's there is there is a there there. so it's incrementally better than what I'm doing and saying, hey, Claude, just don't ever ask me. It's it's incrementally better than that by like at least putting some level of judgment between YOLO and you know. I think it does get to the point where it's like, wait a minute, this this feels like this might be a little. I'm not super confident in saying yes to this. I'm gonna have to ask the human, you know, person to to get involved here. I think there's that aspect already built in. It hasn't asked me yet. I haven't seen a place where it's asked me, but I don't know. I I'm I'm skeptical, but I'm I am optimistic. And they they did do some research as far as far as like we we matched it up against when a human answers the question. Think there's especially Claude code, right? Because it asks to do some like bash commands, like stuff on your machine, like, hey, can I remove your root file system? Yeah, sure, no
Tamara Laine: It yeah.
James Carman: problem. Claude, yeah, go for it. I don't know what that RM minus FR means. I have no idea. but the agent that or whatever system they use, their auto approval thing did, and it was able to say, Yeah, probably don't want to do that. But the humans were, they're just rubber stamping the stuff where that, whereas the agent was actually triggering and and turning it off. So I think there is a there there. I'm with you though. I think when it comes to things that are like I'm not gonna have auto mode just shoop shipping emails out to my clients or posting stuff on my behalf to social media or buying
Tamara Laine: Yeah.
James Carman: things. Yeah, that there's there's some non-recoverable issues that can happen and I'm I I don't want to doing that stuff.
Tamara Laine: Yeah, well like I I just recently set up we we have a new company naming system for all of our files, right? Just because as it grows, you know, too many people are putting things in different places and you know, we had a a wiki at one point and now we we're using no like all the all the different things that go along with that. So
James Carman: Yeah. Yep.
Tamara Laine: I created I created a skill in Claude to just go through and make sure and rename and and name my new file files and all of that so that I don't have to bother with it, right? which is right, one of the real case points for what a good agent does, right? But this this skill kept trying to delete files. And as I and I was cause I was trying to get it more on auto mode, right? I was trying to get it more towards and so like I it was just like now I can't I have to back up, I have to kind of rework this thing because it just keeps trying to be like, this is a duplicate, delete it. And I'm like, it's not it's not a duplicate just because Just because someone doesn't name it right does not mean it's a duplicate.
James Carman: Right, yeah.
Tamara Laine: And, you know, and sometimes, you know, duplicates are fine if you've got them in different places for different reasons. I just there was the whole the whole thing. I was like, no, you can't. Cause like we, you know, we have a data or drive. And so like it was trying to delete things in the data room because it was a duplicate. And I was like, no, stop. You're and just wasn't getting it. So I had to I had to step in. And so I could see where like as simple as that should be, it it turned out to be really something I
James Carman: And the other thing that gives me pause with something like this. So I can go through and I and I I was texting our our producer over the weekend, Ryan, and just like, what is wrong with Opus right now? Have you noticed this? It's like it goes through like these these eras or something where the like the models are just it was just awful. Over the weekend I I I kept having to like undo code that it was writing and and like wow, wait a minute, we decided this 10 minutes ago. What are you doing? Like It was it's been awful for like two weeks. And that scares me because we could you could have exactly like you're talking about, I've got this process dialed in. It's it's identifying the duplicates the right way I won and blah, blah, blah, blah, blah. And then two weeks go by and Claude gets a lobotomy or something. And then and then it starts making really dumb decisions and deleting all my files, which I thought was working perfectly. you know, that goes back to, you know, how do how do you test these systems and all of that and evals and all of all of those fun things. So Crazy. Okay.
Tamara Laine: Yeah, it is it is funny how all of a sudden you're like, why is this not you know, even with like writing tone, right? I've got
James Carman: Mm-hmm.
Tamara Laine: I've got some of my agents really set up to my tone. I've ingested all the podcast transcripts, right? Like it knows how I speak, it knows my it it it knows my tempo, all of that. And then every once in a while it'll send out I'm like, I you know, like I've told you no dashes ever. Like
James Carman: That's the one that gets on my nerves. Yeah, I hate
Tamara Laine: Yeah, I
James Carman: those.
Tamara Laine: you know, and or you know, it goes back to something that's completely predictable and you're just like, What? What happened? And you're, you know, I've got to retrain it again. So yeah, it's weird how just out of the blue it'll do
James Carman: Yeah, and I I've got the, you know, use the Oxford comma. Like you can you can take my Oxford comma out of my cold, comma, dead, comma, and lifeless hand, right? You know what I mean? Like I I am sticking with the Oxford comma. Those sort of things. And then it'll mess them up. Like you you you tell it those the same thing. M-dashes, Oxford com like anyway. I'm with you. It does it it just for some reason it just it just forgets things and doesn't doesn't abide by your rules anymore. Just out of the blue too. mumbling next segment of our show is what we like to call. Our lightning round. And the lightning round, I always have to warn people because you know, everybody hears lightning round and they think, this is going to be fun. It's not fun. This is serious. This is where the answers matter. This is, you know, most people they'll fast forward through all the AI stuff and we're talking about governance and and you know, how do you worry about, you know, making decisions that I can lend people money and that like they're not gonna listen to that mumbo chumbo. They're tuning in for the The where the rubber meets the road, and this that is our lightning round. So the answers are correct or they're not. And I don't know what the answers are. I I don't know. We have an algorithm that we use to score this, and it takes months for it to do all of its number crunching and all of that. we'll get we'll get your score to you eventually. I just I just have to make sure everybody understands because when we say lightning round, that always sounds very kind of whimsical and everything, but it's it's not at all in
Tamara Laine: Serious.
James Carman: any way, shape, or form. It's serious is serious. And you know, this this could be life altering, you know, this sort of score. it's it's it's really it's it's really important business. So I want to just make sure that you're framing your mind around our lightning round and make sure that you're you're ready and you're mentally prepared. Are you ready? Okay. Good. Good, because I think you're gonna see what I mean. real hard hitting stuff. Do you Instagram your food?
Tamara Laine: I used to now no.
James Carman: No. Yeah, my kids are they do that stuff. They yeah, the phones gotta eat first, you know, that kind of stuff. I usually wait so long to eat that I'm I'm I'm gonna kill someone. I give me the food. Don't no, the camera can wait.
Tamara Laine: I kids now, I eat standing up.
James Carman: Yeah, yeah, exactly. In between, you know, volleyball
Tamara Laine: Mm-hmm.
James Carman: lessons and volleyball practice and vol yeah, mostly volleyball for me these days. have you ever
Tamara Laine: No No, I don't think.
James Carman: Good answer. I feel like that's a good I don't know. Like I said, I have no idea.
Tamara Laine: I I'm trying to think. I mean, maybe something from my brother or, you know, my mom's closet, but nothing serious.
James Carman: sourdough or wheat?
Tamara Laine: Sourdough all day every day.
James Carman: Hmm. Now do you make your own? Are you one of those folks that has like your
Tamara Laine: I we've been trying
James Carman: starters?
Tamara Laine: we've been trying to find a starter somewhere because it's like a pet, you gotta feed it. we're still working on
James Carman: Mm-hmm.
Tamara Laine: that. I'm I'm out in the country. We would love to have our own sourdough, but I go buy it from a local farm.
James Carman: Hmm. Yeah, that's the best way to go. Cause I'm with you. Like trying to like like it's almost like having another kid, you know, having this sourdough in your fridge. Like, I got to feed it every day or whatever. Yeah, I can't I can't do that. do you believe in favorite junk food. What are you eating when you're re reaching for some junk food?
Tamara Laine: my goodness, besides if my my junk food is is I would say wine, that's my vice. But if it's a junk food, I would say cheese dip. Cheese dip all day.
James Carman: Are you a are you what kind of cheese dip are you going for? You going like with the beer cheese kind of thing? What are you what where you going? Hmm. Yeah, it's good.
Tamara Laine: Queso, queso dip. Yeah.
James Carman: name a primate besides monkeys and apes.
Tamara Laine: isn't not a whale, isn't like a manatee one? No.
James Carman: I don't know. I d like I said, I don't know the answer to these things.
Tamara Laine: Trying to I'm trying I'm trying to think outside of the box here,
James Carman: We'll see what the algorithm says. I wish I'm like Richard Dawson here. Yeah.
Tamara Laine: All right.
James Carman: Like, or I guess the new guy. What's the what's the new guy on Family Feud? You know, survey says, right? You know, that kind of a thing. I I don't know. We don't have the board up to to to check these questions on on the fly. we can't afford that. We're a low budget operation here, so we'll we'll get there one day.
Tamara Laine: Wait, did you say private an ape? A gorilla would be a a primate, right?
James Carman: yeah, yeah. Primate, yeah, yeah, yeah.
Tamara Laine: A gorilla. I've and I've actually gone gorilla trekking, so g I'm gonna go with gorilla.
James Carman: Have you really? Like you've been up close and personal with him? Wow.
Tamara Laine: In Rwanda.
James Carman: Wow. Man, you've done some cool stuff man. We just have to like stories with Tamara. Like, we're gonna have to do a whole series on this because you've had some really cool stories that we could hear about. was it scary? I would imagine that's pretty scary. Gorillas are they're big.
Tamara Laine: Gorillas, yeah. What turns out is scarier. I I didn't realize that like 10 people a year die on that track. and sometimes it's from like stepping on hornets. There's like killer hornets' nests that you could like step on, or they're all sorts, yeah. I didn't realize that
James Carman: Wow.
Tamara Laine: when I went. Otherwise I think I would have been a little bit more cautious, but it was it was
James Carman: Mm.
Tamara Laine: really an incredible experience.
James Carman: And hippos. Hippos is the thing I'm scared of. I hear they kill a lot of people. Like they're dangerous. Yeah. Yeah.
Tamara Laine: Yeah. Yeah. I I I think it was an AI video, but I saw like a hippo chasing a safari truck.
James Carman: Yeah, yeah. And the one that like in the in the water, it may have been AI too, but it's like hopping out of the water, like leaping out, like chasing the boat on the at like, how in the world has can this huge thing do that? All right. What about this is we've been asking this question a long time for months on the show, and it's it's been an interesting journey for folks. Do you like the smell of gasoline?
Tamara Laine: No.
James Carman: No, just flat out, not ever. Just no. Okay.
Tamara Laine: No. Just no.
James Carman: Do you like the name Charlie for a girl?
Tamara Laine: I could do Charlie for a girl, yeah.
James Carman: Well, I feel like that's a good answer. Have you ever tasted soap?
Tamara Laine: I mean, yeah, I think any kid who grew up in the eighties, someone put soap in their mouth. Someone put a soap bar in your mouth at some point. That's Yeah.
James Carman: Mm-hmm. Yeah. Yeah, we all we all had to do it because Yeah. That's how you learn the boundaries. You you say that F bomb or whatever, and and your the mom's like, no, you didn't, right? Like,
Tamara Laine: Yeah.
James Carman: no, eat this soap for a little bit, and then that'll that'll clean that up for you. Did you go to sleep away summer camp as a kid? Did you? Like
Tamara Laine: Yeah, yeah.
James Carman: as was that like an every summer thing, or was it just once was it a
Tamara Laine: Yeah, it was it was a handful of summers. I went to like
James Carman: Okay.
Tamara Laine: this horseback riding camp when I was younger, which was fun.
James Carman: We sent my oldest to one and and she had never ridden a horse and she went and she rode horse and she just fell in love with it. She loved it. She was like, my gosh, I love horses. So but now she's she doesn't do anything with horses, which but that was kind of a cool cool. Are you a dark chocolate or a milk chocolate kind of person? If you're journaling, this is an interesting because you're a journalist, or a former journalist. If you're journaling, are you gonna reach for paper or are you popping up a computer, iPad, some sort of digital format for your journaling needs? Are you are you going paper? What are doing? She's got paper right in front of her. She picked it up and held it up. That's
Tamara Laine: Paper. Yeah. Paper.
James Carman: I find a lot of people do that.
Tamara Laine: Right in front of me.
James Carman: Yeah. I and I always have a there's nothing in it right now because I was, you know, just leave a page there for me to write things. But I keep a paper journal, right? handy in case I need to write things down. I'm with you.
Tamara Laine: It's really good for the the the memory brain connection.
James Carman: Yeah, that tactile. Yeah, I think, yeah, I'm with you. Okay. And finally, again, we save the kind of the most consequential things. a lot of times we'll we'll like to save that for the last question. what temperature do you set your thermostat at?
Tamara Laine: it's a fight in my house. If I had my way, it would
James Carman: Mm-hmm. It usually is.
Tamara Laine: it would be seventy one to seventy two, but I think it's usually at about sixty eight. Yeah. Yes. Yeah.
James Carman: sixty eight. The husband likes it colder. You like it warmer.
Tamara Laine: I like it more. Yep. Yep.
James Carman: I like it colder too. I do too. I and that's that's interesting because when I ask that question, a lot of folks it's it's well if my you know so and so sets it, it's this. If something if I said it, there's all that's usually the type of answer we get. and our our director, producer, Ryan is saying seventy two is perfect. Well I
Tamara Laine: Yes. So that's right. Seventy one, seventy two. That's right.
James Carman: I think the problem for us, my office is on kind of the main floor of the house. My wife's is upstairs. So we took like a one of the bedrooms or something and converted it into her office. So she's a little warmer up there. So if she'll come down and some days it's too hot, she'll set it, you know, lower, freezing me. And if I'm telling you, I I can't. I need like my my fingers get stiff when it's too cold. Yeah, first world problems, right? Okay. All right.
Tamara Laine: Yes. I will say though at night I like it a lot colder. I do tr I do try to take the Bernie
James Carman: Yeah. yeah, you could get stun.
Tamara Laine: Sanders approach of much colder at night.
James Carman: I thought you were talking about like what he was sitting in that chair, he was all bundled up in that chair. Yeah, that that that Bernie Sanders approach.
Tamara Laine: no, apparently there were stories that like, you know, for because for good sleep, you're supposed to be a little bit like bun like bundled but cold air. And like Bernie Sanders when he was on the campaign trail, like he man it was like something he mandated that his hotel rooms had to be like sixty four or something. I can't remember the exact story. I'm h maybe I'm making it up. Maybe it's fake news.
James Carman: That sounds right.
Tamara Laine: But it sounds right, that he was very much in
James Carman: Yeah. I saw something not long ago, and I'm sure it's yeah, I don't know. Forever since I was a kid, I have had to have a fan while I'm sleeping. And a lot of times I'll have it blow on me. I read something the other day that was like, that's actually bad for you to have a fan blowing on you. It's like for something involved with like your temperature regulation doesn't work right, or I don't remember, but it was like, Freaking me out. Like it can caught like it was causing these like really bad things to happen if you have a fan on your face. I'm like, what have you been doing this fifty years? Are you kidding me?
Tamara Laine: Yeah, I I feel like there's a new study. I'm really trying to, yeah. This is this is me getting off of Instagram because every time I get on Instagram, there's a a new thing that I shouldn't eat or that I've been doing or the workout that, you know, is wrong for my body or all I just I can't do it anymore. You know, and then you look at European, you know, communities and you know, they are just loving life. They're eating whatever they want. They, you know.
James Carman: Mm-hmm.
Tamara Laine: Bread all day, as a
James Carman: Mm-hmm. Yep.
Tamara Laine: staple of life. like all the and you're just like, Well, is it just our stress level? Maybe if it was if we just stopped stressing so much.
James Carman: When you hear people talk about like, yeah, I went to France and I ate everything, desserts and blah, blah, blah, blah, blah, bread every single day and escargot and all this stuff. And I lost weight. Like, what?
Tamara Laine: Yeah.
James Carman: Like, we're doing something wrong here in the States.
Tamara Laine: It's a hundred percent. That was a hard lightning round. Did I pass?
James Carman: Well, like I said, we and we'll get the scores tallied up. We'll have to run it through the system. And we there's a we got to take the transcript and there's a there's a whole embedding process. I I don't even understand all that's going on there. But they tell me it's pretty complex, is what they say. Cause it takes a while. Now I don't know if that just means the people that we have doing that act like we got a crew. We don't have a crew. We have we low budget here. They're just lazy. They don't they'll get around to it. Like I don't know what's going on there. but that's what they tell me. I whatever. but you'll get your score. Yeah.
Tamara Laine: Well I'm very competitive, so I wanna win.
James Carman: yeah. I mean, I feel like I I I like your chances, I do, of of winning the the lightning round. And we will we're we're putting together this badging program where you can actually share it on LinkedIn so you can, you know, brag about your your forward slash score from the lightning round with to others as a credentialing mechanism. Because it's important, as I said. and that'll that'll come in handy somehow. I don't know. But I'm I'm assuming it will. All right. Tell us about anything. This is in closing here. We've we've we've talked about some really interesting stuff. As I said, you are in the thick of it. You're building the kind of systems that everybody wants to try to build, like the things that are making important decisions and doing it wisely and safely with using AI. So you're you're at the forefront. I love it. So thank you for coming on for that. Anything you got coming up? product releases. Are you speaking anywhere? Do you have a a book coming out? Are you planning on starring in any movies where you'll get an Oscar? We'll have to have you come back. Or a t or a Tony. Or a Tony. W either one.
Tamara Laine: that would be delightful. Tony. no, but if anyone ever was trying to make a documentary on something cool and technology, I would be willing to to figure out how to help that. so that's an that's an open offer. because that's just a a passion. If I can if I can put together a couple of passions, I I'm always happy. so what's coming up for me? I mean, I'm the keynote speaker at a couple of conferences
James Carman: Okay.
Tamara Laine: for for credit unions. So I can give you the details on that. One coming up is in New York for the New York Credit Union Association. I'm gonna be talking about hallucinations and legal things that you should look out for. and then what?
James Carman: NYCUA. Is that is that what do they call it that? NYCUA NY Secret Union NYC I mean I don't know if that's the thing. It's just I thought that was an interesting acronym.
Tamara Laine: Yeah. I my podcast is going to be so What the Frack, my podcast that I believe you're gonna be on soon, is going to be
James Carman: I would love to do that.
Tamara Laine: is gonna be broadcasting live from the Boston Startup Week podcast stage. So
James Carman: wow. cool.
Tamara Laine: on September 15th, we're really excited about that. we're gonna have some some real fun on the the Boston stage. And yeah, and I mean our our product is launching the end of this month with our we've got several customers lined up, our first pilot customer. So we
James Carman: Nice.
Tamara Laine: are we are heads down and getting everything wrapped up to be live with them. So that would also mean officially we're live.
James Carman: Gotcha. All right. So that's that's pretty that's big. Is it is that scary?
Tamara Laine: Yeah, it's huge.
James Carman: Is this is it a scary thing or are you excited? Were you more excited than scared?
Tamara Laine: More excited than scared. but I think any any founder should always be, you know, like I n it's never good to be overconfident.
James Carman: Right. Yeah. Yeah. Try to see around the corners of what could be going wrong. Now you
Tamara Laine: Exactly.
James Carman: mentioned like you would help someone with a documentary, you know, think about do you miss it? Do you miss that life of the journalism thing? Is that is that?
Tamara Laine: yeah. I mean, yeah. I I think that there's there was just something so incredibly vibrant about every day in that lifestyle, right? Of just
James Carman: Sure.
Tamara Laine: like kind of voracious learning. and now I'm you know I'm doing the same, but I'm I'm doing it very much in in one place. And I find as a founder I do a lot more digging into to some stuff. So I I think that there's there's some times where I'm like, I would love to be out doing stuff. And I should also say I was really lucky. I gotta do so many incredible things while I was doing journalism. Right? Like you've heard a handful of them. I gotta drive
James Carman: Yeah.
Tamara Laine: a NASCAR. I went skydiving. I
James Carman: my gosh, that's so cool.
Tamara Laine: yeah, I mean I I just I did some things. I went on an there was a snowstorm that like hit the northeast and I went out with an emergency crew all night overnight to just kind of like just see how they're saving lives. And it wasn't a hard hitting story, but it was really impactful because you're out with this EMT team while it was snowing three feet and just like seeing them rescue people.
James Carman: Wow. Man. Now how do you replace that in your life? Like that's pretty that's pretty cool. That's cool stuff.
Tamara Laine: Yeah. Yeah, no. It's
James Carman: Yeah, you can't like you can't like join the HOA board or something to repla you know what I mean? Like, Sally's got her grass too high. No, let's
Tamara Laine: Hey.
James Carman: let's dig in here. Like that's the there's you can't it doesn't compare, right?
Tamara Laine: Yeah, no, I but there's a time and a place for everything, right?
James Carman: Yeah. Yeah. That's really cool. Well, again, as I said, you're you're out there, you're on the on the cutting edge of of of a lot of the things that a lot of people are talking about and you're doing it. So that we appreciate you coming on and sharing insights. how do people get go look up empower and and the what what the frack is that on Apple and all of all of the places?
Tamara Laine: Yeah, what the frack is on all the places. YouTube is where we focus a lot of our attention. But if you want to get in touch with me at LinkedIn, DM me on LinkedIn. That's the fastest way. I'm I'm a little bit behind on my responses this week, but like that's that's where you know you can chat with me. I'd love to connect. and then empower LinkedIn and our website,
James Carman: not E P O D.
Tamara Laine: no, so mpwrai.com.
James Carman: okay. No I thought it was a dot AI. That's everybody's doing the dot
Tamara Laine: Yeah. I wanted
James Carman: AI thing.
Tamara Laine: I wanted the dot AI. Someone someone grabbed that up and it was really expensive. So I just went, all right. yeah, no, someone is definitely just squatting.
James Carman: Are they just squatting on it? I hate that. That's that Ugh I hate that. Well, again, thank you so much for coming on. I really enjoyed the conversation and hope to hear hope to well hope to join you on your podcast. But you know, come back again. We can do more stories about all the cool things you've done in your life. That's that's what I want to hear about. But thank you.
Tamara Laine: Love this conversation. Thanks for having me on.
James Carman: All right, if you'd like to get in touch with us here at the forward slash, drop us a line at theforwardslash@callibrity.com See you next time.
