TRANSCRIPT 🔗
speaker-0: Hello everyone and welcome to another installment of the Building What's Next podcast. I'm your host, Peter Weingard, and this is the podcast where we sit down with leaders from the world's most forward thinking companies for honest, unscripted conversations about AI, transformation, and what's being built in this moment. And I'm also joined today by my co host, Altimetric Global CMO. You know him, you love him, your vendorsing, Sani. Hello and welcome.
speaker-1: Thanks, Peter.
speaker-0: It's great to have here. Our guest today has spent over two decades covering technology, the industry, transformation trends, both as like editor-in-chief in roles such as ZDNet and Solonus Media, to his current position as the editor-in-chief of Constellation Insights at Constellation Research, where he digs into how enterprises are navigating AI software and digital disruption. He's also an adjunct professor at Temple University. And has spent an entire career asking hard questions about where technology is really headed. So please welcome Larry Dignan to the podcast.
speaker-2: Hey, thanks for having me.
speaker-0: It's a pleasure. Thank you for being here. I'm looking forward to this conversation because you're the perfect person. You've you've sat at this perch looking at enterprise technology from different vantage points. And we're sitting in a really interesting place now. a year ago I think we were all talking about, you know, which model should we use enterprises, and now enterprises are focused their conversation about how do we t put this into production and how do we control it. So Sort of with that as background, what do you see as changing in the engineering services?
speaker-2: I think it's gonna be a lot about model routing and also just managing your models as if it's a portfolio. So I I think the services companies, I think enterprises are gonna look to offload a lot of that model management, so to speak, right? Almost like a like you'd buy a mutual fund and they would manage the portfolio. because right now it's you know, people are trying to manage their own stuff and You know, it's it's even brought SAS back from the apocalypse, so to speak. because people are just like, I you know, everybody got burned by tokens like basically four months ago or so. and now they're sort of looking like, Do I really want to do this? And so I think the services firms, I think a lot of it's gonna be about optimization, managing those models, and basically, you know, getting to an outcome that works.
speaker-1: So let me let me follow up on that, Larry. And I know that some of this also sounds familiar because you in fact interviewed our CEO, Raj a a few weeks ago, if I'm not wrong. where we spoke about the whole point of token maxing and FT maxing. and I'm sure, you know, when you you publish a a very extensive piece of research almost every every day. I really look forward to the email that comes over the weekend because it kind of, you know summarizes the whole week for me and and a shout out to you. folks should go and subscribe to what Larry does. if you reflect back on the discussion you had with Raj, and I think he also spoke on similar lines. And I'm sure you've attended other events after that. It was great to see you at Dreamforce as well. And I know you referred to that part as well. What did you take away from that? And I know this was your second time covering Raj. I think first time we were together at the Futures Forum back in March as at Rosewood Sandal.
speaker-2: Yeah, I I think the biggest takeaway from that conversation was the whole maxing facade. What are the two primary optimization traps you see out there right now?
speaker-3: One is token maxing, which which we have seen and the industry has realized and a lot of enterprises are talking about it, you know, the example being Microsoft and Uber, which were publicly talking about how this is an optimization trap for them. And base based on that now I see a trend which I'm framing it called FDE maxing, which is a forward deployed engineers where the
speaker-2: Okay.
speaker-3: where the the frontier labs are Creating companies called deploy calls, which are hiring FTEs and maxing those FTEs in enterprise is not going to get you production grade implementation. You need to think through a lot more from an architecture perspective and how that fits your broadfield reality within an enterprise. So that's the FTE maxing. That's the second optimization time, right?
speaker-2: whether it's token tokens or FDEs, you gotta get to the business outcomes at some point. Yeah. And that's what a lot of that's what a lot of enterprises. the good news is I'm pretty bullish that enterprises will get there. I think we're I think we're entering the grown up era of AI. where I I think we're really gonna have a tail of two markets. You're gonna have your debt bubble AI infrastructure stuff. Which I think's probably gonna blow any minute. I don't think the models are sustainable, the debt loads are sustainable, the whole financing thing, the demand. Like I think that could be an absolute mess. it's just a matter of time. Now, if I knew the time, probably wouldn't be sitting here talking to you guys. but on the other side, enterprise AI has matured a lot. Yeah. and you know, I think think we're getting to a point where enterprises are just getting a lot more savvy about this and you've seen it happen quick. Right? So I mean just the the thing Raj was talking about with the you know FDE maxing and token maxing, it was all just about vanity metrics. Implying that yeah if I do mu enough of this something magical will happen. And everybody got slapped upside the head by their CFO and realized that's not the case.
speaker-0: In in your conversations, where are you seeing AI value showing up in enterprises now? What are the grown up enterprises, what are they doing and what are they seeing?
speaker-2: It's a little of everywhere. I would say a
Load full transcript
year ago it was all productivity based. right now I see a lot of really all sides. what's interesting though is it is going use case by use case, which I kind of I kinda like because I kind of like that process wonk kind of stuff. and they're just kind of snowballing it. And then, you know, if you look at something like Caterpillar. where, you know, they're taking models from NVIDIA and basically, you know, trying to create autonomous vehicles for like construction sites, which is a crazy difficult problem. and then you have a company like Eli Lilly, which they're running their own AI factory for scientific discovery. And, you know, that's about revenue generation. Right. That's about efficiency. And and you know, I saw I saw their lead AI and ML guy at the Core Weave conference last week. And You know, he was, you know, kind of chuckling. He didn't have to worry about token costs because they own it. So we're in this weird time where there's a lot of a lot of stuff that we wrote off for dead is is back in f back in favor, notably on prem. Right? Because you're you're literally looking at this and going, Hey, I don't I don't want to get hosed on token costs that I you know, like that that model's great if you are Chat GPT or a hyperscaler. For the rest of us, it's just a cost.
speaker-1: And also Larry, to follow up on Peter's question, you mentioned the two names that you took also belong to two very different sectors, right? One is big big manufacturing and and the other is pharma. in your view, and maybe a two part question, how do you see this playing out when you look at specific verticals or domains? And then in your view, is there any sector that you think is front ending this because you know, some background for our viewers. Ultimatry plays big time into the regulatory industry. So you know a lot of our customers are belonging to BFS and and pharma sector. So any view that you can give from that. And then of course I have a follow up question on a very exciting event, Ray and you guys are in New York. So that's coming after this.
speaker-2: Yeah. So leading up to now, I mean been tracking this probably three years now for the most part. what what I've seen repeatedly is that companies that are regulated seem to be doing the best. And there's a reason for that. They're regulated. They have to have their data ducks in a row. Which when you look at AI, you know, most of this stuff is the biggest challenge is what's your data look like? Can AI read it? Can you share it? Do you have a data lake? All that good stuff. and the, you know, the JP Morgan's, the, you know, the regulated industries and financial services, they've been, you know, I mean, I would venture to say somebody like JP Morgan Chase is is a good two years ahead of everybody. Just in how they've integrated in the business, you know, they did their center of excellence stuff. And then now it's just part of the mix, right? It's it's not this separate thing. It's just part of the business. and that's where I think everybody's gonna get to, but I do see the regulated industries doing pretty well on that front. And I think the big reason is it it's not the regulation. It's it's the regulation has forced your data hand. So you know your customer, you actually have to you know, you you had to do stuff just for the regulations, and it's worked out pretty well. you know, healthcare has some interesting use cases. I don't think that's you know, necessarily gone gone crazy yet. but, you know, pharma, that's legit, right? you know, and and the Lily person was talking about the AI factory and they have the biggest AI factory in pharma, but that's gonna last about a year because I think Bristol Meyer Squibb and somebody else is doing something. so they have their own arms race going on. And, you know, I the on prem thing's interesting just because, you know, the the enterprises that have like you know, real data that's their their moat and their engine. you know, they want their own infrastructure because they don't want to you know, and they'll build they'll build on Nematron or some open model or develop their own models, but their you know, their secret sauce is their data and they're not really willing to put that somewhere. They want control of it all.
speaker-1: Is is there a role, Larry, that like yes, as a as a industry or as a sector, yes, but there is still a difference between leaders and Lagarth, right? Do you think there's a role that leadership or how bold are companies willing to be, does those nuances and culture impact how they perform here as well?
speaker-2: culture's huge. top down is huge. you know, I think you you were at our futures forum event in March and you know the running the running gag was that the CO is really the chief AI officer. Yes. largely because they're the ones that are getting pelted by questions from the board. you know, in March that was totally true. Now You know, I think we got some at AI Forum that event we had. Yeah. Seems like it's been taken back a little bit. but there's you know, it it is a it is a board level issue.
speaker-1: Is it s sorry to interrupt you there because you said something very interesting there because between March and now, is it because the layer below the CEO is is going up and taking that responsibility? Like, why do you think it's coming down the chain?
speaker-2: I think it's coming down the chain because it's one thing if your CEO, you know, wants a AI pilot, you can they get that across pretty easily. but when you actually start to deploy, there's little things like budgets, AI costs, are we locked into a model already? Did we just create tech debt over the last three months? You know, all these kind of things. so I think it's You know, I think that's part of the you know, if we look, you know, let's let's rewind the clock to like twenty twenty four. Remember there was all these enterprises, they were like, Yeah, we're standardized on open AI. And that was
a great press release for Chat GPT and the enterprise. T Mobile comes to mind. There's a few of right? They we probably all remember the press releases and you're like, Ooh, well, that was a board level push. And a CEO push. And it sounded great at the time. But now you're like, well, there's multiple models. I don't want just closed models. I want open models. And and you know what I mean? So you almost even what we're creating today, any architecture that's created this minute, it's probably going to be legacy in two months. And then you have this decision where do I re architect? Do I just rewrite everything? it's moving that fast. So, you know, the the art here is I think the art here is creating an architecture that you don't get locked in, and the other art is knowing what not to do. Right? Like sometimes the best moves you make are the ones you don't do.
speaker-1: In action also action. Yeah. All right.
speaker-2: You know, so you have to be very focused.
speaker-1: And then sorry Peter, I have a I have a chain of thought and I and I wanna I want to cover that. Now the participants in future forum were board and CEOs. Yes. While in the AI forum that you did in New York City recently had about two hundred plus practitioners. You know, these people were actually the chief AI officers of companies and people who are leading the charter. And then I know that I think that there is already out on when you're doing your next future forum. So your prediction or what do you think we will hear when we re reach that event next? And do you think there's any takeaways that you want to share in terms of how you see into the future when we do the that event?
speaker-2: If I'm gonna break out my crystal ball, I would say twenty twenty-seven's a year of government governance.
speaker-1: Governance.
speaker-2: because I you know, cybersecurity has everybody freaked out. I don't think we've even started to see I don't want to call them rogue agents because they're just agents that you didn't think through enough what not to do. Right? When it's it's very interesting framing where, you know, they're Open AI and even Anthropic to some degree are like, okay, this agent hopped a sandbox and you know, NVIDIA comes out with their open agent safety thing like a week ago, a week and a half ago. And, you know, the whole idea is this is an engineering problem. That's part hardware, part software. And you need to you need observability and see what these agents are doing from the get go. But I do think like we're we're probably gonna run into a liability issue at some point. You know, I know we're all into the, well, we need regulation and there's this huge back and forth, but at the end of the day, who's on the hook? And there's always a human on the hook at these things. You can't sue an AI agent yet. I mean, maybe Muse gets sued at some point. but you're looking at, you know, there's liability, there's governance, there's the whole cybersecurity issue, and I think
enterprises are really gonna try to button this thing down. Especially before especially when they set these automated processes running and all that. They're gonna have
speaker-1: Interesting. one final question and Peter, I'll let you you know, pick it up from there. so governance is the main point, Larry, that you say. In the again, coming back to my my industry question, is there a specific industry that you think is ready or do you think will be the pioneer which will set the benchmark when it comes to how to manage the governance aspect here?
speaker-2: I would have to go to financial services. Because they they are tech companies. they are also, you know, they are used to being pelted with attacks every second of every day. You know, for years, like before AI, whatever. and they're they're big targets. So, you know, I think when you're looking at governance and managing agents and all that, I I would have to go with financial services because I just think they're They're just stronger, like just even from a base of talent to, you know, people have known AI and ML for years. I mean, remember, all the all the trading that happens today are algorithms. So it's not like not like financial services has to, you know, suddenly understand what an algorithm is. so I expect them to have the government's game. I d you know, and then you're and then you can have like the industrial AI folks, the manufacturing places, anybody doing that stuff. is gonna be, you know, on the on the governance train. So I mean, it's gonna be most industries. you know, the industry I worry about the most is probably ours. It's tech. We're a little loosey goosey with all this stuff, right? you know, when you just look at these personal AI agents and what they want from you and all your data, I mean, I'm like, what? I mean, people give up data all the time. So You know, I guess that's good, but I I expect some epic disaster.
speaker-1: Not Peter?
speaker-0: Well, I want to get back to the the sort of the AI factory model, which you mentioned earlier, you said some of the companies are setting up AI factories. I also recall you've written in the past that you thought, you know, this AI factory model will eventually wane. what would it be replaced with and why do you think it'll it'll wane?
speaker-2: Think we have a couple options. one, you know, AI factory in the enterprise is gonna have some select use cases. So if you look at a company like Lily, right, they're trying to find breakthrough drugs, you can justify that expense. others are gonna wind up using, you know, a bit of the cloud and bit hosting. But the part where I look at the AI factory and the whole, you know, let's let's let's just take up a whole state and build a data center. It is so inelegant. Like it it just like you look at all the distributed compute we have running around, that if you can harness that somehow in a peer-to-peer thing, leverage leverage the compute on your device which isn't being used or whatever. You know, maybe there's some transaction where you rent space from you know, but there's gotta be some way to use distributed compute better for inference. Instead of, you know, taking up all of Texas with the data center.
speaker-0: Right, right. So sort of the AI equivalent of companies building their own replacements for out of the box software.
speaker-2: Yeah, and I I just see the infrastructure should be more distributed. And if you just think of just think of the telcos, right? The AT and Ts, the Verizons. What's our big problem? It's it's land, it's power, it's you know, approvals. They have all that. And they have all these central offices that are laying around with, you know, a bunch of compute that could easily be swapped out for GPUs or whatever. I'm just not I'm just not sold on this, let's build the biggest damn mainframe we can find. maybe we need that for training, but for inference, I gotta think there's a better way to do it.
speaker-0: That's fantastic. I never actually thought of a all these companies already have sort of distributed computing architecture built as part of their previous business.
speaker-2: Yeah, I and I think that's why Nvedia has the DGX spark. Right? Like there's you know, and and for companies, they gotta look into it, right? Because say you say you tune and do all the stuff with the models locally. I mean those those mach those desktop machines have some serious horsepower. Like they're they're mini data centered to themselves. And you string a few of those together and then you don't have the token costs, right? You can It doesn't go production until you you've optimized this thing to the hilt for cost or efficiency or answers or whatever. I think we'll see more of that. Cause I'm just not, you know, and I don't if you just look at the power requirements and you know, it just doesn't it does not make a damn bit of sense. Yeah. So there's gotta be something else. I just don't think in the US we're engineered to think that way. Mm-hmm. Right. I think we're it's it's meathead AI is what I call it. Right. It's just we're like, ooh, let's do the biggest, baddest thing we can do and I'm not sure that's the most elegant solution.
speaker-0: W when we first started chatting, you said, I you know, it's the fourth quarter already. The year's nearly over. If we're to look back at some of the thoughts you had maybe at the beginning of the year, what of the what of the sort of predictions you had or assumptions you had at the beginning of the year are playing out the way you thought and what are things that have taken a different turn and not worked out the way we thought they might have?
speaker-2: well my AI factory thing has not played out yet at all. you know, I was thinking, well, it is earnings season coming up, so we'll we'll see how this works out. But I do think, you know, there there might be some slowdown in terms of the capital spending going into AI factories and stuff, largely because
interest rates are going up. The financing's gonna get harder. So I think that's something to watch. I it's not gonna crash. I just think it's all it has to do with slow. the other thing I talked about was the IPOs of OpenAI and Anthropic. Anthropic looks like it'll get out the door. Open AI not so much. but my working theory was that was gonna be the peak. I think we've already seen peak anthropic. To be honest, I I think Peak Anthropic came with all the token maxing and all that stuff. And you know, you see the reports of people cutting back on cloud usage because of cost and all that. that's gonna be a very interesting IPO to watch. And I think we may already been at the peak of that. But remember, all these things can be true at the same time. You can have the best of times in enterprise AI and the rest of you know, the rest of the household names, maybe not so great. Or certainly turbulence. so the jury's still out on that one. we'll we'll see how it plays out. the SAS pocalyp seems to be over. but you know when I say over, it's basically, you know, most of those stocks have gotten back to flat. So, you know, it we'll we'll see how it goes. but you know, we'll we'll ultimately see those knock-on effects for various companies. So You know, Meta launched the Muse AI agent and you could see some financial services stuff took a hit. Because you know, Planet Fitness took a hit. Because the thinking was you would use an AI agent to cut subscriptions that you know you may or may not use it reuse, right? Like a gym. we'll see how that plays out. But you know, initially it looks like a crazy overreaction, and then over time we'll see how it works. I think SAS we're still sorting out winners and losers.
speaker-1: I think it's interesting how correlations can work out because as GLP drugs came out, I think there was a thing about grocery stocks taking a hit because people will consume less. So Right interesting dynamic. in fact, you know, I I I I was at the last week was an interesting week because I was at the World Economic Forum's Center for AI AI excellence. There was a meeting that happened and as expected there was a lot of talk about sovereignty. Right. And and the interesting part is as you double click, the question was more about resilience. You know, can the business keep running if a model gets restricted or if a cloud goes down? you know, I I would love your your view on that. That's that's one question. And second is one interesting thing, there was a survey that that was done and and it seemed like eighty-four percent of the participating companies They have not redesigned a single job around AI. you know, it was fascinating for me to me to hear that, given the kind of representation that was there. so on one side, you know, AI sovereignty and resilience, and on the other side, the fact that everything is a bolt on. What's what's your view on these two data points?
speaker-2: resilience, I think, is gonna be a big theme. I mean, all you
gotta do is look at the Middle East and data centers there that took a hit. so I think that's that's an issue. I don't know if it's gonna be front and center till something really happens. but I think you know there's ways to manage around it, right? It's you're multi-cloud, you'll have some of your own infrastructure. what you'll wind up doing is Find your most valuable models and make sure they're very resilient and can hold up. but you know, we're we're like we're in a crazy volatile world. Yeah. So, you know, I think the whole AI resilience conversation is mixed up with sovereignty, data sovereignty, AI sovereignty, and just can you take a punch and rewire yourself to do it, right? and then as far as the jobs go, I'm not sure I would be rewriting jobs right now because AI is moving too fast to really know what you need. And I think that's why we saw like you know we saw some job cuts, but frankly, I think they were all just they just used AI as an excuse. They got too bloated during COVID and didn't cut. but that said, I think it's a case of you know, you paused hiring to figure out what you need. So we had the great pause. now, you know, now companies seem to be hiring more entry level and all that. but it's hard to rewire a company that's already scaled. You know, like AI Forum, there were a few smaller companies and you know, one person was up there talking about, you know, they had a team of eight rewriting something, then they're like, Well, that's too many. I only need six, I only need four. And now they're like, Is two too big? Because you're orchestrating a bunch of AI agents. so the small companies, they can punch way above their weights. So I think they've organized jobs around AI. the larger ones, you know, it just remains to be seen. But I I do think companies are being rewritten and rewired as we speak. and I I do you know, I do suspect in the next year or so we're gonna have a massive company that's rewritten and transformed and done all this stuff for AI. And that that might set a template for others. but I do think, you know, the work will change. organizations will change. They'll probably be organized around work as opposed to functions. you know, I I think where we overshoot is we expect this all to happen yesterday. And there's just there's a lot of change management, there's a lot of process, there's a lot of leg there's so many moving parts here. Yeah. I think it's gonna be a lot like digital transformation. Right. Which which is interesting because if you know, like I was I checked out McDonald's Investor Day and they had a lot of AI talk. They're using Google distributed cloud in the restaurants and all this kind of stuff. and you know, their their point was, yes, we've been For the last seven years we've been doing this digital transformation stuff, getting the CEO said they're going from data puddles to data lakes, you know, and all this stuff. but you know, that's that's the precursor for any of this stuff. So if you're a company that's lagging on this transformation effort and the data efforts from, you know, if you're behind the eight ball on that, you know, how how fast can you actually rewire a job?
speaker-1: I think this this reminds me again of, you know, quoting Raj when we were launching Alti Labs, you know, which is an in house lab where we do a lot of experiments and working through models and mixture of models, etc. he spoke about the fact that you will need even more engineering rigor. You know, because a lot of what you spoke about needs that rigor, needs that ability to be able to follow that and this will clearly differentiate companies that really have that engineering strength. versus the ones who don't. So I feel very confident in terms of all the points that you're raising in my internal discussions and discussions that we're having with our customers, a lot of these are front and center. So really great to see how we are resonating on these points, Larry.
speaker-2: Thank you. And you're also gonna have to pick between deterministic workflows and what the models can do, right? I mean, I plus I mean honestly, I mean, in most of my experience, you know, the models are great interns. Some of them are really bad interns, some of them are good interns. but you can't just let them run.
speaker-1: Yeah.
speaker-2: And until you get there. But to get there you're gonna need a lot of deterministic, you're gonna need lot of engineering to make that happen. I think not to mention process know how.
speaker-1: Yeah. And and I think the other thing that I feel is being a practitioner. What I also feel very confident is that when I look around myself in terms of my leadership and a lot of people that are on the team, these folks come from the industry. you know, these are folks who've been practicing. And maybe I wanna put Peter on spot because, you know, as as as our global head of brand, I think Peter, you should also share with our audience in terms of how you are using AI. And and I know that we were debating about how you have all these interns that work for you now. Do you wanna give any view to our audience in terms of we did a big brand refresh recently on how did you use AI?
speaker-0: Yeah, well we couldn't have done it without AI, Larry. We you know, we we integrated S L K software into the company. we you know, ten thousand employees and we're in what, sixteen countries, is that right, Griffinder? Mm. And imagine you know, even for a young company, all the legacy web pages and documents and things floating about that had to be rebranded and just we did it in I think six weeks. Everything from ID badges to you know, the the office designs, laptops, and if not for the power of AI to read through hundreds of web pages and hundreds of documents to say which ones are on brand, which ones are talking about old stuff, which ones don't need my attention. You know, it we couldn't have done it with a team as small and as agile as ours. And we use
it constantly. you know, Claw Design, we have it, it can ingest Figma files. and keep things on brand, developing you know, all kinds of assets. Yeah, it's a wonderful tool, but like you said, it's not infinite cognition or infinite judgment, it's infinite interns, and needs constant observation and management.
speaker-2: Yeah, I think the biggest challenge I have is sometimes it thinks too much.
speaker-0: Well certainly writes too much sometimes.
speaker-2: Well, it writes too much, but you're also like, you know, you can ask it something and it'll be like thinking, pulling this skill, pulling that skill, and you're like just asked a simple question. You know what I mean? Like it's shouldn't be this difficult. So I don't I don't think we're there yet. you know, part of my part of it might have been my prompt engineering wasn't buttoned down enough. That's entirely possible too. but you have to be really specif like it it it misses that intuition thing.
speaker-0: Absolutely. Yes. And we've all made the mistake of using the more sophisticated model for the simple question. Right.
speaker-1: Let me that that reminds me of a of a fun talk that I was having with a very senior leader who said that you know, I used to be very direct in my emails and I used to write to people, better do your job and and write something like a two liner and I said that now what I do is I go to a tool, a AI tool, and I and I write that the AI tool tells me well, you don't write like that and it writes the same message in eighteen pages. And guess what? The team who gets that message, they put Nobody has the time to read those eighteen pages. So they put those eighteen pages back into the AI tool, which tells them you better do your job. So yeah. It's it's an interesting spiral. talking about, you know, moving back to you also have a big event coming up, C C E the annual get together at at Half Moon Bay, with Ray and and the entire team. Any Any preview you can give, I'm assuming we'll release this episode before the event, Peter. Let's try to do that. any insight you can give on folks that are coming, what is what is the big topic and let's let's not talk about AI, apart from AI.
speaker-2: I think well in three bullets. One, it's networking. two, you walk out of there with some big ideas and a whole lot of takeaways. and it it's just a way to basically set up your year going forward, right? So topics aside, you'll you'll come away learning a lot. you'll come away learning, you know, you'll you'll meet you'll network with a lot of your peers. And it just sets up the ear well. I mean that's how I use it, right? I I use it to basically like, all right, so what what themes do you need to watch? Whatever. And yeah, you know, I mean, honestly, by my third day, my brain hurts a little because I've soaked in too much. starts falling out my ears or whatever. but, you know, it's all good stuff that I can come back to, you know, in the the later two months of the year and just sort of, you know, get your own thinking. You know. You can use as a jump off point to do your own thinking.
speaker-1: Well, I can I have done it for ten plus years, I guess, now, and I can totally echo your sentiment on all the three points. and and but I also feel that it's the way in which you you plan the event that yes, if it gets too much, there is also that opportunity to go network with people. What I love is that somebody who's I think eighty percent to ninety percent of the crowd is on stage at some point in time. So that's surrounded by people who will be on stage. And so that that makes the whole discussion very, very impactful. And I and I think to your point, the number of actionable insights that I walk away with, where I definitely always come back and call my team on the Monday after the event, and I want to share with them and then I want to figure out what actions are we gonna take to implement whatever we've learned at the at the forum. So kudos to you and the entire team for setting this up.
speaker-2: Yeah. And and like I said, I sit there and just try to get as many takeaways as I can.
speaker-1: Do you want to remind the audience on the dates when when the event is?
speaker-2: I believe it's the twenty sixth.
speaker-1: Starting the twenty sixth of October at the Awesome.
speaker-2: October, yeah, half moon day.
speaker-0: Fantastic. Well, Larry, this has been a great conversation, Gravinder. Thank you so much. this is so much fun. let's just leave it with I wanna ask you, Larry, what should the listeners who are some people you suggest they follow besides yourself, to to see what's happening in the world of enterprise AI and and the tech industry in general? Who are people you follow and like?
speaker-2: I will read I mean part of this is just like skew that way anyway, but I'll read a lot of AI bears. I'll I'll read I'll read the folks that are just kind of poking around the balance sheets, finding things that you know, and and the problem with being bearish is you always sound logical, but if your timing's wrong, you're hosed. so I'll look a lot of that. I will read a lot of earnings call transcripts. So
speaker-1: Yeah.
speaker-2: So I mean that's kind of kind of my hobby in an odd way. and you know, I always look at industries that aren't tech. Right. So I I wanna see what how the Pepsi CEO is thinking about something. I wanna see what Caterpillar's thinking. I wanna see, you know, how Walmart's thinking about its AI agent, Sparky. which is apparently just killing it in terms of driving cart size and all that. So I'm looking at those industry folks. And I'm looking at what the CEOs say on record. In a in an earnings call. Yeah, 'cause they they say a lot of stuff.
speaker-1: And I I'm reading this book. I just met this it's from a Stanford professor, Melissa Valentine. And fascinating. I'm actually gonna recommend to Wala and Ray to bring her on Disrupt TV if not already done. And again it's fascinating how this talks about teaming with AI along with the with the smaller team. Again, I'll not disclose the full details. People should go and read the book. But yeah, let's for I I love your perspective, Larry. I never thought the AI BR spot is very very interesting for me. I think I'm gonna I gotta insight into how how you're thinking. So this wasn't
speaker-0: Yeah, you give us a lot of thinking read about it.
speaker-2: Part of just my cynical nature, like I'm reading nineteen twenty nine too. which is a nice reproduction of how that stuff and it it's a fascinating read. There's
speaker-0: I j
speaker-2: so many historical things in there that you're like, this is where this business school name came from, right? Like it's it's it's pretty fascinating to see how it's all out all
speaker-0: And talk this weekend actually, I give a presentation.
speaker-2: Yeah.
speaker-0: and it's there are a lot of parallels between nineteen twenty nine and today.
speaker-2: Yeah, and as a journalist I I just like the legwork. It's
speaker-0: Yeah.
speaker-2: it's a zillion documents. The you know, the footnotes is probably a fourth of the book. so I appreciate the legwork basically. Okay. But it
speaker-1: That's the right thing list, yeah.
speaker-0: Yeah, another time
speaker-2: It's a fun read.
speaker-0: it's fantastic. Thank you everybody, and we'll see you again soon on the next episode of Altametrics Building. What's next?
speaker-1: Thank you. Take care. Bye.
speaker-2: All right.