Vibe Coding for the Enterprise: How AI is Rewiring the Modern Web Stack | Guy Yalif (Webflow)

WEBVTT
00:00:00.239 --> 00:00:16.079
Webflow last summer, 8%, so not one, but eight percent of our self-served signups came from LLM sourced traffic, referred traffic, uh uh where it was a direct referral, not counting the Halo effect on direct nav and branded search, ignoring that.
00:00:16.079 --> 00:00:26.160
That traffic converted six times better, not six percent, six X better than unbranded organic search, which I think is the fair comparison.
00:00:26.160 --> 00:00:28.719
And what I've heard from CMOs, eight is a little high.
00:00:28.719 --> 00:00:31.920
I've heard like two, three, four, five growing rapidly.
00:00:31.920 --> 00:00:34.159
The 6x, that was not an outlier at all.
00:00:34.159 --> 00:00:37.759
SEMrush and HRES found 4x to 23x.
00:00:37.759 --> 00:00:45.920
Uh, and so it is like higher converting uh traffic because they've had the conversation and are further down the funnel.
00:00:45.920 --> 00:00:51.359
I have heard some people say, oh, well, therefore, your site doesn't need to have upper funnel content that anymore on it.
00:00:51.359 --> 00:00:58.880
And I could agree more because the LLMs are looking to your site also as a data source, and you need to be able to tell that story.
00:00:59.280 --> 00:01:00.159
Hey, what's going on, guys?
00:01:00.159 --> 00:01:02.399
This is Kevin Kerner with Tech Marketing Rewired.
00:01:02.399 --> 00:01:04.239
Uh, this session was a really fun one for me.
00:01:04.239 --> 00:01:07.840
I talked to Guy Yaleff, who is the chief evangelist of Webflow.
00:01:07.840 --> 00:01:09.040
Yeah, Webflow.
00:01:09.040 --> 00:01:16.879
And uh as Guy found out over the last few weeks of talking to him, I'm a big Webflow fan.
00:01:16.879 --> 00:01:29.359
And uh I was really excited to talk to him about all the changes that are happening in CMS and marketing sites and micro apps and all the things that are at the top of every person's agenda right now in marketing.
00:01:29.359 --> 00:01:42.319
So Guy and I talked a lot about the dynamic between vibe coding and what's happening with micro apps and everyone having access to all these powerful tools and being an enterprise that needs to stay secure.
00:01:42.319 --> 00:02:00.640
He also talked about um how CMOs are actually in the tools, playing around with some of these technologies now more than ever, and how they're organizing their team to take advantage of this new capability while also staying secure and um building for the enterprise.
00:02:00.640 --> 00:02:02.239
So it's really great conversation.
00:02:02.239 --> 00:02:04.799
Guy is a great human being and really easy to talk to.
00:02:04.799 --> 00:02:06.879
And we could have we could have gone on forever.
00:02:06.879 --> 00:02:14.240
But um, this was a really insightful conversation from someone who talks to CMOs all the time about this stuff.
00:02:14.240 --> 00:02:19.280
Before we get started, though, I wanted to mention this podcast is uh sponsored by my company, Mighty and True.
00:02:19.280 --> 00:02:28.560
Mighty and True is a growth marketing agency that works with tech brands to help them mature their growth strategies and also help them uh move along their AI and automation initiatives.
00:02:28.560 --> 00:02:33.360
So we do a 45-minute blueprints session with companies as a way to get started.
00:02:33.360 --> 00:02:47.840
Um that blueprint will allow you to identify those gaps in your strategy where we might be able to improve things, but also you can run AI and automation experiments inside those growth marketing efforts to uh help mature your use of these new technologies.
00:02:47.840 --> 00:02:57.520
So uh if you're interested uh and want to reach out for a blueprint or reach out about the podcast, you can reach us at www.mightyandrue.com.
00:02:57.520 --> 00:02:59.039
Okay, that's about it.
00:02:59.039 --> 00:03:01.360
Can't wait to for you guys to hear from Guy.
00:03:01.360 --> 00:03:03.039
This is Tech Marketing Rewired.
00:03:03.039 --> 00:03:04.960
Let's get to it.
00:03:04.960 --> 00:03:14.000
Guy, welcome to the podcast.
00:03:14.000 --> 00:03:15.120
Great to have you.
00:03:15.120 --> 00:03:16.639
Thanks for having me, Kevin.
00:03:16.639 --> 00:03:17.919
Great to connect again.
00:03:17.919 --> 00:03:20.960
I am so excited to have you on.
00:03:20.960 --> 00:03:28.479
Uh, you know, I'm I'm the I'm a giant uh Webflow fanboy here, and I reached out to you to to uh talk to you about what was going on.
00:03:28.479 --> 00:03:32.479
And it it's like the dream, you must have the dream job right now at Webflow.
00:03:32.479 --> 00:03:45.680
I was just looking through the uh the LinkedIn Webflow uh profile, and the I saw the Vidoso acquisition, which is asset generation, I saw the One Trust thing, which is consent and user preferences.
00:03:45.680 --> 00:03:53.360
You've got component canvas launched now, you've got native in cursor, so 10 skills and MCP server.
00:03:53.360 --> 00:03:57.120
It's just MCP and yeah.
00:03:57.120 --> 00:03:59.919
How do you what do you even do when you wake up in the morning?
00:03:59.919 --> 00:04:03.280
Are you like, what is gonna what do I get to play with next?
00:04:03.280 --> 00:04:04.800
What's what could happen next?
00:04:04.800 --> 00:04:05.680
It's really great.
00:04:06.000 --> 00:04:16.240
It it is awesome to partner with an EPD team that is constantly shipping stuff and enabling people to just run.
00:04:16.240 --> 00:04:28.560
Uh and we were the first CMS to support MCP, and now like the things you can go do directly out of your LLM of choice, it's it's really a lot of fun.
00:04:28.560 --> 00:04:30.560
I'm grateful for the partnership in the Fanboy Dome.
00:04:30.560 --> 00:04:32.000
I happen to fanboy as well.
00:04:32.000 --> 00:04:34.879
I'm like, hey, this is really empowering.
00:04:34.879 --> 00:04:46.639
And I know we'll talk about it more, but the concept of doing that empowerment in a safe, brand safe, security, privacy-safe way.
00:04:46.639 --> 00:04:48.000
I like that.
00:04:48.560 --> 00:04:52.399
Yeah, let's just let's jump into that because that's really that's where my head is at right now.
00:04:52.399 --> 00:05:02.720
Like we're just doing a little bit of discussion on the free call, and um, it's kind of interesting to talk about the Vibe Code thing and how creating apps from scratch and overnight and all that kind of stuff.
00:05:02.720 --> 00:05:17.360
But there is this dual reality going on right now of people that can create pretty much anything, but then there's the other side of it where you have enterprises that are, you know, need to be secure and keys and credentials and access.
00:05:17.360 --> 00:05:23.759
And there's a real threat out there from bad actors to to get into things that you shouldn't be in.
00:05:23.759 --> 00:05:29.759
So I wonder from your perspective on the Web Plus side, like what do you, how do you, how are you guys wrestling with that?
00:05:29.759 --> 00:05:33.519
Con, I don't know if it's conflict, but those just those two dimensions.
00:05:34.319 --> 00:06:01.519
Embracing it fully, saying that uh with the mental model of yes, we absolutely need to enable each of us to create, call it vibe coding, call it agentic development code, but that you know uh mere mortals who don't know how to code natively, or those who do, but to be able to do these things at scale in plain English guiding, the prompt to prototype, it's mind-blowing.
00:06:01.519 --> 00:06:02.879
It's truly magical.
00:06:02.879 --> 00:06:07.439
It is excellent, but prompt to production is a whole other thing entirely.
00:06:07.439 --> 00:06:27.120
I I like I hope it doesn't happen, but I would bet a pretty penny that some well-intentioned, I'll pick on marketers, because I've been a marketer for a while, somewhere is going to get fired one day because they vibe a lead gen magnet and they do a great job of it and it gathers PII and then they get hacked the next day.
00:06:27.120 --> 00:06:37.839
I I think having a uh safe playground where you've got security, you've got governance, you've got privacy.
00:06:37.839 --> 00:06:49.040
And then if you're enterprise, you're probably also going to want like roles and permissions and workflows and reviews, but like you can't skip the security, governance, privacy at a business of any size.
00:06:49.040 --> 00:07:02.639
Uh and and so I like the idea of being able to play safely, where you're like really empowering people to go bring these visions to life, but you're doing it in a secure privacy governance safety way.
00:07:02.639 --> 00:07:14.079
At Webflow, we then think, in my humble opinion, in a differentiated way, of also making it brand safe where the playground you're playing in, you're gonna stick in your design system.
00:07:14.079 --> 00:07:18.160
You can have the great ideas without it breaking the brand.
00:07:18.160 --> 00:07:20.560
And that's pretty cool.
00:07:20.800 --> 00:07:22.879
Yeah, yeah, I it is.
00:07:22.879 --> 00:07:38.240
And and the thing I'm the thing I'm playing around with now that I see a lot of is the micro app, the adjacent apps that are outside of the web flow environment, and creates there's so much creativity that you can make out there.
00:07:38.240 --> 00:07:46.079
It's how do you manage these self-created tools that might plug into an ecosystem?
00:07:46.079 --> 00:07:55.120
Like that self-hosting side, running something up on Vercel or something, having some uh headless container.
00:07:55.120 --> 00:07:56.879
You can do that really easily.
00:07:56.879 --> 00:08:06.800
But um, I wonder how you guys think of the surrounding ecosystem that might be vibe coded apps that need now need to plug into your enterprise grade CMFs.
00:08:07.439 --> 00:08:15.279
A few different ways, all of which follow those same principles we were talking about before, where you get to build the way you would like.
00:08:15.279 --> 00:08:19.680
By the way, you can vibe code an app on Webflow.
00:08:19.680 --> 00:08:30.399
You can host an app on Webflow using Webflow Cloud and then Webflow Code Components and Webflow, you can literally just genetically go create things, and there's plenty more coming there.
00:08:30.399 --> 00:08:43.759
But let's say whether you built it on Webflow and hosted on Webflow Cloud or any other place, you then can plug it in and have it live in this brand safe place, these shared guardrails.
00:08:43.759 --> 00:08:50.159
Uh, I think when you're doing it in, I think the ebb and flow of the microservices more broadly, right?
00:08:50.159 --> 00:08:53.600
You and I have seen the suites are the answer.
00:08:53.600 --> 00:08:54.879
The point solutions are the answer.
00:08:54.879 --> 00:09:00.240
The suites are the right, and that ebbs and flows over the course of better parts of a decade.
00:09:00.240 --> 00:09:00.720
Okay.
00:09:00.720 --> 00:09:15.440
Now I think we will find, I personally believe, and I the webflow too, that there is uh incremental value and goodness in having shared context where that context isn't is is your brand voice, your style guide.
00:09:15.440 --> 00:09:26.879
It's also all those other things we were talking about, those guardrails that make it safe to do what you're doing, so that you can go really fast and be really creative and be safe at the same time.
00:09:27.759 --> 00:09:31.039
Yeah, I yeah, I totally, I totally understand what you're saying.
00:09:31.039 --> 00:09:41.840
From a um leadership perspective, to back up a little, you work with a lot of CMOs and CIOs, and I think you were at the conference recently where you had a lot of CMOs at SEO conference.
00:09:41.840 --> 00:09:43.840
What's on their minds around this stuff?
00:09:43.840 --> 00:09:51.120
Like how I wonder how aggressive do you see marketers wanting to be, or CMOs specifically, like leadership.
00:09:51.120 --> 00:09:56.480
And then what do you see on the other side when you get the web dev, uh uh secops, all those three?
00:09:56.480 --> 00:10:02.080
Like, what are the what are the two different dynamics that you hear going on between those two in an enterprise?
00:10:02.879 --> 00:10:05.759
Very happy to share to put it in context.
00:10:05.759 --> 00:10:14.320
Uh I was uh AI coding aerospace engineer, uh, then became a CMO, because you know that's a straight line, right?
00:10:14.320 --> 00:10:16.879
Aerospace engineer to CMO, ha ha ha ha.
00:10:16.879 --> 00:10:28.000
I spent uh like um 15 years in ad tech, and then 10 years ago was co-founder CEO of an AI website personalization company that Webflow bought a couple years ago.
00:10:28.000 --> 00:10:33.200
Uh and the reason I have those conversations that you were describing is because okay, I was a CMO as a CEO.
00:10:33.200 --> 00:10:34.480
Webflow acquired us.
00:10:34.480 --> 00:10:35.919
Webflow has both of those.
00:10:35.919 --> 00:10:40.240
So we made up this job called Chief Evangelist, where I'm out in market talking to CMOs all the time.
00:10:40.240 --> 00:10:42.879
And so I have hundreds of conversations each quarter.
00:10:42.879 --> 00:10:56.799
And then the direct answer to your question, having given that context, is yeah, they are working to adopt as quickly as they can, but that it varies wildly.
00:10:56.799 --> 00:11:06.559
There are places I feel like cut me off is not where you want to go, but like there's a bit of an AI fluency, AI maturity model we can have in mind.
00:11:06.559 --> 00:11:10.159
And the answer to where the CMOs are depends on that for their org.
00:11:10.159 --> 00:11:16.960
I feel like, you know, the the the initial part is like privacy, security, legal, saying it's okay to use these tools.
00:11:16.960 --> 00:11:28.799
There are a lot of enterprises like Fortune 500, where it's still not okay to use a bunch of these things, where folks uh uh are still saying, hey, um it's gonna train on my data.
00:11:28.799 --> 00:11:31.679
And and you know, yeah, haven't gotten to that point.
00:11:31.679 --> 00:11:32.159
Okay.
00:11:32.159 --> 00:11:39.120
Then you enable people to play level two, level three, you enable them to play, but pointed at business goals.
00:11:39.120 --> 00:11:57.279
Uh and hopefully somewhere between level two and level three, people have their own like aha moments where they then, to what you and I were talking about in the pre-call, are up for going through the easy and the frustrating parts that enable them to have agents running for you know uh many hours at a time.
00:11:57.279 --> 00:12:01.600
I would humbly submit when you go from there, then you automate parts of your job.
00:12:01.600 --> 00:12:09.120
Many CMOs are hoping that would happen, not because I haven't heard CMOs say, oh, hey, I'm looking to cut a bunch of people.
00:12:09.120 --> 00:12:13.519
I've heard them say, like, let's take Yeah.
00:12:13.519 --> 00:12:14.559
Same here.
00:12:14.559 --> 00:12:15.120
Yep.
00:12:15.120 --> 00:12:18.799
That let's take the work that's less enjoyable, less fulfilling.
00:12:18.799 --> 00:12:24.080
Let's have people use their superpowers, use the do the things that they became marketers to do.
00:12:24.080 --> 00:12:26.639
And so four level four lets you do that.
00:12:26.639 --> 00:12:42.960
Level five in this made-up rubric is automate things to superhuman levels uh that you couldn't do before, like gather all the things before you meet with somebody for a meeting in a way that like you never would have found the time to do before.
00:12:42.960 --> 00:12:48.960
And level six is the stuff you read headlines about, but I feel like there's a lot of distance between here and there.
00:12:48.960 --> 00:12:54.639
And that's truly re-architecting your work, your org, how you think about things.
00:12:54.639 --> 00:12:58.960
And so I feel like there's a lot of articles up here.
00:12:58.960 --> 00:13:01.919
I feel like there's a lot of dependencies down here.
00:13:01.919 --> 00:13:07.840
And I feel like many of the CMOs are in that middle of, okay, let me go find the champions.
00:13:07.840 --> 00:13:09.600
Let me go try this very directly.
00:13:09.600 --> 00:13:14.320
I've seen more hands-on direct work than in anything I've ever seen before in my career.
00:13:14.320 --> 00:13:18.960
I see you nodding too, because I feel like it's it's experiential, right?
00:13:18.960 --> 00:13:22.559
To like actually have you're you're like, wait, what are the degrees of freedom I have?
00:13:22.559 --> 00:13:23.360
What are the use cases?
00:13:23.360 --> 00:13:24.720
What can this thing do?
00:13:24.720 --> 00:13:35.279
And it feels a little bit, it's not quite like at all like parenting, where like, oh, you can't really describe what it's like to be a parent until you are one.
00:13:35.279 --> 00:13:36.879
You know this better than I do.
00:13:36.879 --> 00:13:40.879
Um uh, but but uh you need to experience this.
00:13:40.879 --> 00:13:42.000
We both know it.
00:13:42.000 --> 00:13:42.799
We both know it.
00:13:42.799 --> 00:13:48.000
We both do, but but two and eight are two very different things to scope.
00:13:48.000 --> 00:13:51.759
Um and then on on the Go ahead, go ahead.
00:13:51.759 --> 00:13:52.000
Go ahead.
00:13:52.000 --> 00:13:53.039
No, do you keep going?
00:13:53.039 --> 00:14:05.039
And to answer the second part of your question, uh, I feel like on the whole, the technical folks are thinking, um, hey, great, like we can go build this whole thing.
00:14:05.039 --> 00:14:09.519
We'll give you total control because we can go run so quickly.
00:14:09.519 --> 00:14:11.679
Uh this is one story arc.
00:14:11.679 --> 00:14:12.799
It's obviously not the only one.
00:14:12.799 --> 00:14:14.559
Every company and person are different.
00:14:14.559 --> 00:14:20.000
But I feel like the the it sounds super appealing up front.
00:14:20.000 --> 00:14:24.799
And then you vibe code something, but then you get to day two where you're like, wait, now I need to manage this thing.
00:14:24.799 --> 00:14:26.320
Wait, now I want to run a test.
00:14:26.320 --> 00:14:39.120
Wait, now I want to tweak this individual word, and you create a dependency on engineering that puts us as marketers back where we were, which is like unempowered, not because engineering doesn't care.
00:14:39.120 --> 00:14:39.679
They truly care.
00:14:39.679 --> 00:14:41.039
That's why they built this in the first place.
00:14:41.039 --> 00:14:43.919
They wanted to make this better, but because they're busy building the core product.
00:14:43.919 --> 00:14:45.200
They were like, hey, I want to run a test.
00:14:45.200 --> 00:14:47.200
And they're like, you don't even know if this is gonna work.
00:14:47.440 --> 00:14:47.759
Yeah.
00:14:47.840 --> 00:14:48.080
Yeah.
00:14:48.080 --> 00:14:50.080
I'm gonna go build a core product.
00:14:50.080 --> 00:14:56.480
And so that contradiction, I think, um uh folks are increasingly gonna run into.
00:14:56.480 --> 00:14:58.080
Sorry, that was a super long answer.
00:14:58.159 --> 00:14:58.879
How does that look like that?
00:14:58.879 --> 00:15:00.639
Really very, very good.
00:15:00.639 --> 00:15:09.440
Um, and I think it gets me thinking that you have this at the CMO level, you have this pressure to be in the things, doing the things like never before.
00:15:09.440 --> 00:15:13.600
It's like, and I I've talked just this past week, I've talked to two CMOs.
00:15:13.600 --> 00:15:18.720
One was a current CMO, and the other was a CMO that had left her job and was something else.
00:15:18.720 --> 00:15:26.960
And they're both have this intense desire and pressure, really pressure, to get in the pools and start doing the things.
00:15:26.960 --> 00:15:38.720
Uh like you never heard a CMO want to get into like Aliqua or you know, Marketo or something and do the things, um, or an you know an ESP platform or anything else.
00:15:38.720 --> 00:15:48.720
But for some reason, and I guess it's because we're all empowered to get in these things, there is a big push to actually do the work, to create the things, to experiment.
00:15:48.720 --> 00:16:01.120
And I do think it makes them better leaders if they are able to understand the requests, the wildest requests I get from people are the ones where they don't have a grip on what the stuff can do and the dangers of it.
00:16:01.120 --> 00:16:10.399
And they just want to, you know, do all the things all at once, but it's not backed up with any um experience or understanding of what the tools are.
00:16:10.399 --> 00:16:14.399
Um so I I was I completely agree with you.
00:16:15.200 --> 00:16:33.600
I uh to your point about the two CMOs, I think some of the most AI knowledgeable, AI advanced people I talk to now are the ones that have been out of out of a full-time job for an extended period of time because they are finding the hours that everybody else is like, boy, I wish I found the hours.
00:16:33.600 --> 00:16:37.200
They are finding the hours and are really doing breakthrough things.
00:16:37.200 --> 00:16:45.759
And to your point about yeah, as a CMO, chances are pretty good you were not honkering to get into the you were picking on send email.
00:16:45.759 --> 00:16:46.159
Great.
00:16:46.159 --> 00:16:48.159
You know, you pick on any one of the systems, right?
00:16:48.159 --> 00:17:08.079
Like that's totally, I feel like uh I think it's stating the obvious uh something you would agree with too, but like the pace of change is so fast, like measured in months and weeks, and that it's truly breaking constraints that we have been optimizing around for a long time.
00:17:08.079 --> 00:17:10.960
And so the game we're playing changes, right?
00:17:10.960 --> 00:17:13.359
That the constraints we're optimizing against change.
00:17:13.359 --> 00:17:20.000
And then I feel like the other part of it is um culture is one of the very few things that flows from the top.
00:17:20.000 --> 00:17:34.319
And so I feel like there's a healthy dose of inspiring people on their teams who are any, you know, anywhere between like, oh, I'm all in, this is the future, to I'm scared I'm gonna lose my job, to I don't understand the words that just came out of your mouth.
00:17:34.319 --> 00:17:39.279
Like I feel stupid asking about it and all the above.
00:17:39.680 --> 00:17:41.119
Yeah, for sure.
00:17:41.119 --> 00:17:43.200
Um that lands for you?
00:17:43.200 --> 00:17:44.799
Yes, absolutely.
00:17:44.799 --> 00:17:50.880
Um on the flip side, there's the CIO point of view, or maybe the web dev team or CISO.
00:17:50.880 --> 00:17:57.839
Um, if I'm a CMO, what do I need to know about that side of the the organization?
00:17:57.839 --> 00:18:00.480
Like how are this, how's the CISO feel about this?
00:18:00.480 --> 00:18:03.039
How's the web dev team feel about all this bot coding?
00:18:03.279 --> 00:18:06.400
How can I manage them better as a CMO?
00:18:10.079 --> 00:18:22.480
Um and and you know I feel like at a pattern matching stereotype level, so by definition, does not apply to everybody.
00:18:22.480 --> 00:18:39.680
Um I think part of the reason engineers are finding so much value in the LLMs is because you can get more out of them with thinking that is taught when you're trying to be an engineer.
00:18:39.680 --> 00:18:44.559
Taking big problems, breaking them down into pieces, thinking about systems and how they talk to each other.
00:18:44.559 --> 00:18:54.640
And I feel like again, at a stereotype level, most uh go-to-market people, marketer sellers, like that wasn't part of uh art history degree.
00:18:54.640 --> 00:18:56.559
That wasn't part of some of the things they were taught.
00:18:56.559 --> 00:19:05.440
And and but you get more out of these systems when you uh are thinking about them that way or like breaking down jobs to be done.
00:19:05.440 --> 00:19:09.599
Product managers, again, stereotype, are very, very good at doing that.
00:19:09.599 --> 00:19:16.240
Most you you you go to a uh stereotypical AE and say, hey, what are the jobs to be done you're doing every day?
00:19:16.240 --> 00:19:19.920
And and like the the the it's just not the way they've been trained.
00:19:19.920 --> 00:19:31.599
And so I feel like there is part of the AI fluency that is raising their fluency and thinking that way and also lowering the bar so they don't have to think that way.
00:19:31.599 --> 00:19:39.039
I feel like some of the best AI I'm seeing is like it's it's in context in a workflow they're familiar with before you change everything.
00:19:39.039 --> 00:19:44.000
It's context aware so it knows their brand if that's relevant or knows their tech or so on.
00:19:44.000 --> 00:19:47.359
And then it's asking people questions they're familiar with answering.
00:19:47.359 --> 00:19:52.559
And so for the engineer and the marketer, those are going to be two really different things.
00:19:52.559 --> 00:19:53.920
Does that land for you?
00:19:54.160 --> 00:19:54.880
Yes, yeah, yeah.
00:19:54.880 --> 00:20:08.720
And one of the great hacks I can think of is if you are a marketer and you need to talk to the engineering team or you need to make a change, um, ask the LLM to give you some context of how they might feel, like the systems thinking that an engineer might be thinking.
00:20:08.720 --> 00:20:17.279
If you're a marketer, you can very easily ask the LLM to help you have that conversation in a way that, you know, helps enable, you know, so you sort of meet in the middle.
00:20:17.279 --> 00:20:20.880
So there's some context for the developer or the engineer of the web teams.
00:20:20.880 --> 00:20:22.799
Because the LLMs are a great thought partner.
00:20:22.799 --> 00:20:25.119
We don't all have to be experts anymore.
00:20:25.119 --> 00:20:32.240
You know, we've got we've got a great thought partner that can help us be semi-experts or at least relate to each other easier.
00:20:32.240 --> 00:20:40.000
Um, I want to pivot a little to the um to another big topic here that we've talked in with pre-meeting about, which I think is you were at this conference.
00:20:40.000 --> 00:20:48.079
Um, I can't remember the name of the conference, but um it was all around AEO, DEO and AEO or AEO um best practices.
00:20:48.079 --> 00:21:00.480
And you you know, you heard a lot of people, a lot of uh senior marketers talking about what they're doing in terms of um AI search and readiness for AI search.
00:21:00.480 --> 00:21:03.920
I wonder if you could just give some of your thoughts on the things you heard.
00:21:03.920 --> 00:21:06.480
What was the name of the conference that you spoke at?
00:21:06.960 --> 00:21:07.920
There were a bunch of them.
00:21:07.920 --> 00:21:14.880
I had the privilege of co-hosting AEO Conf with the CEOs of Graphite and AirOps.
00:21:14.880 --> 00:21:19.759
And it was it was it was a great day.
00:21:19.759 --> 00:21:25.359
First, the teams did a great job physically setting it up, but additionally, it's such a rapidly evolving space.
00:21:25.359 --> 00:21:36.319
We're all learning together, and it was awesome to hear both of them to speak with them, to speak with Twilio, OpenAI, Reddit, uh SQL, and and uh a bunch of others.
00:21:36.319 --> 00:21:43.519
Um uh and uh on the and and the thing that was really interesting was okay, it was solely CMOs and marketing leaders.
00:21:43.519 --> 00:21:47.039
Uh it was the rare conference where we actually said no outside of that.
00:21:47.039 --> 00:21:49.119
If we were very fortunate, the interest is so high.
00:21:49.119 --> 00:21:54.480
There were 1200 people on the wait list for this because the interest in the topic is so high.
00:21:54.480 --> 00:21:55.039
Okay.
00:21:55.039 --> 00:22:07.440
The CMOs uh to it connect to your earlier question, there was very Real interest in actionable playbook level discussion for the CMO, which is different than the practitioner's playbook, right?
00:22:07.440 --> 00:22:14.480
Like, here are the ways to manage your team, here are the ways to guide your team, here are the ways to evaluate your team, here are some of the early plays to go run.
00:22:14.480 --> 00:22:19.039
Not like, here's the, you know, the way to execute this thing, right?
00:22:19.039 --> 00:22:20.400
That their team would want to know.
00:22:20.400 --> 00:22:28.960
There were very there was real actionability desire, if I could make up that word, you know, from the CMOs around what do I do next?
00:22:28.960 --> 00:22:32.079
Um, and we we talked about a bunch of them.
00:22:32.079 --> 00:22:55.039
Um we we invited CMOs to think about AEO as answer, sorry, answer engine optimization, which is really the same goals we had with the SEO, where you want to show up in an LLM or show up in search, you want to attract people to your site, drive more revenue and pipeline that way.
00:22:55.039 --> 00:22:55.519
Okay.
00:22:55.519 --> 00:23:01.039
Sometimes people call it GEO, AIEO, we're calling it AEO, and happy to talk about that if that's interesting.
00:23:01.039 --> 00:23:07.279
But um, we invited people to think about AEO as close to what good SEO always should have been.
00:23:07.279 --> 00:23:08.720
Not identical, but close.
00:23:08.720 --> 00:23:13.839
And that your SEO and AEO teams, your SEO and AEO agencies, they're one and the same.
00:23:13.839 --> 00:23:17.519
It's the right background for AEO to do SEO.
00:23:17.519 --> 00:23:17.920
Okay.
00:23:17.920 --> 00:23:22.400
And then we invited people to think about it in four different categories.
00:23:22.400 --> 00:23:24.559
Is this interesting to keep going on, or should we change topics?
00:23:24.559 --> 00:23:26.480
Yeah, no, no, no, let's do it.
00:23:26.480 --> 00:23:32.960
The four categories were content, technical authority, and measurement.
00:23:32.960 --> 00:23:45.359
And we invited people to think about it that way because in content, you're going from this world where, you know, to get organic traffic, you were thinking about okay, what topics do I want to be known for and show up for?
00:23:45.359 --> 00:23:48.319
And those are represented as clusters of keywords.
00:23:48.319 --> 00:23:57.759
Okay, well, now those same topics are represented as clusters of questions that people are asking because in Google search time, we all learn to speak search.
00:23:57.759 --> 00:23:59.759
Our average career is four words long.
00:23:59.759 --> 00:24:04.319
In LLM time, we're speaking much closer to plain English.
00:24:04.319 --> 00:24:08.160
Average career is 23 words long and growing, and you have a full conversation.
00:24:08.160 --> 00:24:11.279
So you get these richer, deeper questions.
00:24:11.279 --> 00:24:11.839
Okay.
00:24:11.839 --> 00:24:31.440
And then we invited people to think, hey, you know the questions your prospects are asking, but if you're looking for them, like go run an LLM over your sales call recordings, over your sales and support emails, go ask the LLM, go look at Google's people also asked, like all of those are good fodder, and then answer them throughout the funnel.
00:24:31.440 --> 00:24:34.480
What's a good CMS?
00:24:34.480 --> 00:24:36.880
How does Webflow compare to X, Y, and Z?
00:24:36.880 --> 00:24:38.640
How long will it take to implement this?
00:24:38.640 --> 00:24:41.839
Right, those are throughout the funnel, and then answer them in depth.
00:24:41.839 --> 00:24:57.839
Because again, to pick on Webflow with a totally made-up example, the concept in SEO time of answering um how do I personalize on my CMS for mobile versus desktop in New York for finance?
00:24:57.839 --> 00:24:58.720
I made that up.
00:24:58.720 --> 00:25:04.400
It's a silly example, but like that level of specificity that far down the tail in SEO land, useless.
00:25:04.400 --> 00:25:08.799
In AEO land, maybe if you got a bunch of the fundamentals right.
00:25:08.799 --> 00:25:11.200
And then you want to make the content personalized.
00:25:11.200 --> 00:25:23.039
In technical, you go from on-page SEO and all the thoughtful things we did around keywords to the single biggest thing you can do, serve it up on a silver platter.
00:25:23.039 --> 00:25:30.400
And by those made-up words, I mean make it easy for the LLMs to understand the structure and meaning of your content.
00:25:30.400 --> 00:25:41.440
So it's easy for them to ingest and decide in their trade-off between how expensive is it to crawl you and uh how good is your content, make it cheap for them to consume you.
00:25:41.440 --> 00:25:49.759
So you can do that in the content itself with bulleted lists, with uh descriptive headlines, with table of contents, with FAQs.
00:25:49.759 --> 00:26:03.759
And you can do it technically in the metadata, which everyone knows is stuff the machine sees, but the people don't, with things like SEO metadata, accessibility metadata, schema metadata, which uh is is incredibly valuable to add.
00:26:03.759 --> 00:26:05.680
Um and I'll briefly go through the other two.
00:26:05.680 --> 00:26:12.880
In authority, you go from backlinks, which still matter, to because LLMs are looking for consensus.
00:26:12.880 --> 00:26:15.759
Plain text dimensions on other sites matter a lot.
00:26:15.759 --> 00:26:19.039
So brand is back because it creates consistency.
00:26:19.039 --> 00:26:22.559
PR is back because you show up in a bunch of different places.
00:26:22.559 --> 00:26:28.640
Uh, and lastly, in measurement, you go from keyword ranking to okay, I care.
00:26:28.640 --> 00:26:30.160
Did I rank for these keywords?
00:26:30.160 --> 00:26:32.559
Now I care, did I show up for these questions?
00:26:32.559 --> 00:26:34.400
I care to be in the answer to.
00:26:34.400 --> 00:26:37.039
And then you mature to what was my share of voice?
00:26:37.039 --> 00:26:42.880
And then three ideas that just don't exist in search, which is because Google was using our words.
00:26:42.880 --> 00:26:45.359
Was the sentiment positive?
00:26:45.359 --> 00:26:47.759
Was the message pull through good?
00:26:47.759 --> 00:26:49.279
And was it accurate?
00:26:49.279 --> 00:26:51.920
Or did it confidently get it wrong about me?
00:26:51.920 --> 00:26:53.759
Again, sorry, long-winded answer.
00:26:53.759 --> 00:26:55.519
How did that land for you?
00:26:55.599 --> 00:26:56.480
What does that trigger for you?
00:26:56.480 --> 00:26:57.599
Really good stuff.
00:26:57.599 --> 00:27:06.880
Yeah, but it's it is it's really interesting to hear you say all those things, but they they are kind of fundamental SEO um best practices, really.
00:27:06.880 --> 00:27:07.839
They're the same thing.
00:27:07.839 --> 00:27:09.680
They're just they're just turned a little different.
00:27:09.680 --> 00:27:18.240
The idea that the the queries are the search terms, you know, long tail search, let's call it, um, the search terms, they're real long now.
00:27:18.240 --> 00:27:19.759
It's such an interesting concept.
00:27:19.759 --> 00:27:31.839
And the thing that you the thing that is uh that comes to mind is that uh that it now you can use AI to actually go get some of those longer tail ideas for you.
00:27:31.839 --> 00:27:34.400
You don't have to like sit and come up with them.
00:27:34.400 --> 00:27:38.400
Um the AI can actually enable you thinking of some of these things.
00:27:38.400 --> 00:27:40.000
It's really it's really fascinating.
00:27:40.000 --> 00:27:46.079
I uh interviewed um the SEO lead for Procor a couple podcasts ago.
00:27:46.079 --> 00:27:55.759
He made the point that only 1% of traffic for most companies is coming through LLMs, but you had a significant portion of your total funnel coming through traditional search.
00:27:55.759 --> 00:28:02.799
So you don't want to over-index on it, I guess, but you really want to get ready for it, I would guess.
00:28:02.799 --> 00:28:09.039
I mean, under a lot of the CMOs, there's all this pressure to be there in AEO, which I totally understand.
00:28:09.039 --> 00:28:13.119
But most of the funnel is coming from you know your traditional stuff right now.
00:28:13.839 --> 00:28:16.880
I agree with that SEO leader pro core a lot.
00:28:16.880 --> 00:28:26.240
I've seen slightly different numbers, but the principle's the same, that it's small, growing exponentially, and really high converting.
00:28:26.240 --> 00:28:42.480
At Webflow last summer, 8%, so not one, but eight percent of our self-served signups came from LLM sourced traffic, referred traffic, uh uh where it was a direct referral, not counting the Halo effect on direct nav and branded search, ignoring that.
00:28:42.480 --> 00:28:52.559
That traffic converted six times better, not six percent, six X better than unbranded organic search, which I think is a fair comparison.
00:28:52.559 --> 00:28:55.119
And what I've heard from CMOs, eight is a little high.
00:28:55.119 --> 00:28:58.319
I've heard like two, three, four, five growing rapidly.
00:28:58.319 --> 00:29:00.559
The 6x, that was not an outlier at all.
00:29:00.559 --> 00:29:04.160
SEMrush and HRS found 4x to 23x.
00:29:04.160 --> 00:29:12.319
Uh, and so it is like higher converting uh traffic because they've had the conversation and are further down the funnel.
00:29:12.319 --> 00:29:17.759
I have heard some people say, oh, well, therefore, your site doesn't need to have upper funnel content anymore on it.
00:29:17.759 --> 00:29:25.359
And I couldn't have disagreed more because the LLMs are looking to your site also as a data source, and you need to be able to tell that story.
00:29:25.359 --> 00:29:27.279
And not everyone is further down the funnel.
00:29:27.279 --> 00:29:42.079
But to your point, I would look at total LLM source traffic and invest in AEO, but don't throw out the baby with the bathwater, if that's the right metaphor for this, because 90 plus percent is still coming from search in other places.
00:29:43.279 --> 00:29:50.720
Is there anything that um Webflow is able to do since you control that whole, you know, the ecosystem uh in inside your platform?
00:29:50.720 --> 00:30:00.480
Is there anything that you're doing technically that will help um companies adopt or become more ready for the AEO?
00:30:00.480 --> 00:30:07.920
Like there are any technical parts of what you can do or could or could do in the future that help rank get ready for AEO.
00:30:08.559 --> 00:30:13.119
Um, we happen to have invested a ton in this and will continue to do so.
00:30:13.119 --> 00:30:21.119
If we focus just on the technical category and not the other three, um we make it really easy to serve things up on a silver platter.
00:30:21.119 --> 00:30:32.880
So we're the only place I know of now that will uh look at your content, pick which ones of the hundreds of schemas, like product page, company page, event page, bio page.
00:30:32.880 --> 00:30:39.039
Pick the right one, apply your content into it uh for that page or for a whole CMS collection.
00:30:39.039 --> 00:30:53.440
So auto-generate schema, auto-generate sitemap XML, auto-generate metadata based on images, uh, auto-generate SEO metadata based on the page, um, create code that is clean and semantic.
00:30:53.440 --> 00:30:58.559
And the LLMs like semantic code so that you're telling them this is an H1.
00:30:58.559 --> 00:31:01.039
Then here are logically structured H2, H3s.
00:31:01.039 --> 00:31:03.359
Other places might do it as like just a bunch of divs.
00:31:03.359 --> 00:31:05.680
That doesn't give the LLMs as much signal.
00:31:05.680 --> 00:31:20.559
Mobile responsive by default, 300 plus global data centers, super fast, super secure, updated security automatically, all the things that help an LLM consume the content and think, hey, this is trustworthy.
00:31:20.880 --> 00:31:21.200
Yeah.
00:31:21.200 --> 00:31:22.559
Okay, that makes a lot of sense.
00:31:22.559 --> 00:31:25.519
And I we're a Webflow customer, so I see it firsthand.
00:31:25.519 --> 00:31:26.799
It's really good stuff.
00:31:26.799 --> 00:31:35.200
Um, if from the conference, from the CMO discussions you had around all this stuff, how are they thinking about organizing around this need for AEO?
00:31:35.200 --> 00:31:39.920
Did you hear anything about just the team, who does this, what type of skill sets?
00:31:39.920 --> 00:31:45.279
CMOs are probably scratching their head, or but they probably understand, but they're like, guys, a lot of stuff to do.
00:31:45.279 --> 00:31:49.039
Like, what do I need to enable on my team to make this stuff work?
00:31:49.599 --> 00:31:57.119
The majority I have heard say, which in my humble opinion is the right answer, it's my SEO team or my SEO agency.
00:31:57.119 --> 00:31:58.079
I don't need to make a change.
00:31:58.079 --> 00:31:59.759
I might need to buy them a new tool.
00:31:59.759 --> 00:32:03.680
The listening tools for did I show up in LLM answers?
00:32:03.680 --> 00:32:12.400
Uh, some Russian HRFs, which won the keyword ranking war, you know, are also doing this, and there are a whole bunch of others.
00:32:12.400 --> 00:32:15.119
Okay, but generally it's the same folks.
00:32:15.119 --> 00:32:23.519
I I think many are softly evaluating the team a little bit differently because like organic traffic for many is flat or even down some.
00:32:23.519 --> 00:32:32.319
But I think we're in a temporary and temporary multi-year time where, like the early days of search, things are changing so rapidly.
00:32:32.319 --> 00:32:39.680
The reason I think we asked everybody to have organic traffic going up and to the right for so long is because things were pretty steady state.
00:32:39.680 --> 00:32:43.680
Well, now we've got this two sets of traffic, right?
00:32:43.680 --> 00:32:49.759
This like uh some of the search traffic is going down a little bit, but then you've got this much higher converting LLM traffic.
00:32:49.759 --> 00:32:56.000
And so I think some of the more thoughtful CMOs are evaluating it separately and not beating up their teams because traffic's going down some.
00:32:56.000 --> 00:32:59.680
And they're looking at a little later funnel, like what were total leads or conversions like?
00:32:59.680 --> 00:33:04.000
I have heard a couple of stories that I think are uh really scary.
00:33:04.000 --> 00:33:17.599
The it it the caricature of it was totally unvalidated, so I do not know if it was real, but somebody I trust said to me, um, hey, I heard a Fortune 500 CMO said, you know what, I want to go do AEO now.
00:33:17.599 --> 00:33:18.079
I'm all in.
00:33:18.079 --> 00:33:20.799
I fired his or her entire SEO team.
00:33:20.799 --> 00:33:21.359
Yeah.
00:33:21.359 --> 00:33:25.359
I think most of what we do is shades of gray.
00:33:25.359 --> 00:33:26.240
This is black and white.
00:33:26.240 --> 00:33:27.759
Like that is the wrong thing to do.
00:33:27.759 --> 00:33:30.079
That is not what somebody should be doing in this case.
00:33:30.319 --> 00:33:30.480
Yeah.
00:33:30.480 --> 00:33:36.240
Yeah, shades of gray is the perfect way to put it because it's just it's tweaking what you already know about SEO.
00:33:36.240 --> 00:33:47.440
You couldn't have a really good SEO person on your team, it's not a very big leap for them to understand this stuff, both technically and process and if educating the rest of your team.
00:33:47.440 --> 00:33:55.839
But if you take them out and try to be something that's really hard crazy with your site, that's just not that's not a good idea at this point.
00:33:55.839 --> 00:33:58.880
It's it's it's a small bit of the traffic, it's very important.
00:33:58.880 --> 00:34:00.160
It's gonna change a bunch.
00:34:00.160 --> 00:34:01.200
I'm really interested in the state.
00:34:01.359 --> 00:34:03.920
Also, if you were to go ahead, sorry.
00:34:04.319 --> 00:34:18.960
I I'm really interested in agent to agent, uh the evolution to now agent to agent web, where a lot of the stuff that gets served up on the web over time will be agents talking to each other and then serving up.
00:34:18.960 --> 00:34:35.920
It's kind of like I don't know if I mentioned to you before on the pre-call, it's kind of the resend uh email story where if you ask an agent, if you ask, if you go into the cloud or chat GPT or something and you asked about Nemo platforms and you're at least stabbing an email, it will recommend a tool called ReSend.
00:34:35.920 --> 00:34:42.400
And ReSend has structured their web such that the agents find them very easily.
00:34:42.400 --> 00:34:53.679
Um, there's a really interesting story from the guy he launched in that he launched reset, and he started seeing he's getting a lot of traffic from the LLMs, and so he leaned into it all the way.
00:34:53.679 --> 00:34:58.079
And it's just he, you know, he basically built the business till he's uh AI.
00:34:58.079 --> 00:35:01.440
Basically, he sells to the AI agent.
00:35:01.440 --> 00:35:07.360
Um, there's another tool that we use for social media called Late, which is the same thing.
00:35:07.360 --> 00:35:13.840
It's like Late is a social media platform for developers, and it just looks like a developer platform.
00:35:13.840 --> 00:35:17.920
It's got the developer docs and all the things in an agent, which is love.
00:35:17.920 --> 00:35:25.199
So I'm interested in that transition to agent-to-agent communications.
00:35:25.199 --> 00:35:26.639
What do you think about that?
00:35:26.639 --> 00:35:28.159
Are you seeing the same thing?
00:35:28.719 --> 00:35:32.159
Uh I think it is super interesting what you shared.
00:35:32.159 --> 00:35:36.960
Uh I'll just do the one half step back on the SEO AEO teams.
00:35:36.960 --> 00:35:43.360
If you were to define the ideal AEO background to hire for, it'd be an SEO background.
00:35:43.760 --> 00:35:44.079
All right.
00:35:44.239 --> 00:35:45.280
Just to say it out loud.
00:35:45.280 --> 00:35:47.360
To go to the point you were making now.
00:35:47.360 --> 00:35:54.639
Uh we very much see websites as serving two audiences.
00:35:54.639 --> 00:35:57.440
I'll say now, but they have been for a long time.
00:35:57.440 --> 00:36:03.360
All right, websites need to create these visually stunning, emotionally evocative, engaging storytelling for humans.
00:36:03.360 --> 00:36:03.920
Okay.
00:36:03.920 --> 00:36:07.920
They need to then efficiently communicate similar info to machines.
00:36:07.920 --> 00:36:10.400
We did that with SEO metadata.
00:36:10.400 --> 00:36:13.920
We did that with Facebook Open Graph for those of us that are B2C.
00:36:13.920 --> 00:36:18.000
Schema is now the next one, but like that is part of the job.
00:36:18.000 --> 00:36:25.679
Um I'm with you that like the the agent to agent, the A to A, uh, that will be a very big thing.
00:36:25.679 --> 00:36:34.639
I'm individually, not Webflow, super fired up by Google's creation of WebMCP.
00:36:34.639 --> 00:36:41.599
I think it'll enable a lot of additional sites, even ones that don't have a lot of functionality on them, to be functionally accessible to agents.
00:36:41.599 --> 00:36:42.800
Again, opinion of one.
00:36:42.800 --> 00:36:44.400
Very curious where that goes.
00:36:44.400 --> 00:36:46.880
So the resend and late examples.
00:36:46.880 --> 00:36:49.280
Oh my gosh, that is so true.
00:36:49.280 --> 00:36:53.760
You can have an impact tomorrow on AEO by making changes.
00:36:53.760 --> 00:36:58.880
And I don't mean metaphorically tomorrow, because the LLMs are running searches all the time.
00:36:58.880 --> 00:37:00.719
I'll give one example.
00:37:00.719 --> 00:37:15.920
We, and by we, I mean Vivian Huang, who runs our SEO AEO team, um, took our top six product pages, added FAQs to the bottom of them, mirroring the data that was up above, but not structures FAQ, added schema to those FAQs.
00:37:15.920 --> 00:37:22.719
In two weeks, 57% of the incremental citations Webflow.com got were for those six pages.
00:37:22.719 --> 00:37:24.400
That's a super interesting set.
00:37:24.400 --> 00:37:27.039
When you realize Webflow.com has a quarter of a million pages on it.
00:37:27.039 --> 00:37:31.360
Those six got 57%, and they got 24% more traffic in two weeks.
00:37:31.360 --> 00:37:37.679
How hard would we work in search time to get the backlinks, to drive the authority, to drive that much traffic?
00:37:37.679 --> 00:37:39.679
And so you really can have a big impact.
00:37:39.679 --> 00:37:46.400
And I think there is a maybe it's arbitrage, it's certainly early mover because it's a bit of a new medium.
00:37:46.400 --> 00:37:51.119
And those that haven't yet will get there later, but they'll do more work.
00:37:51.119 --> 00:37:54.559
Those that do earlier, again to your point, moderated.
00:37:54.559 --> 00:38:06.320
It's not a hundred-zero ignore search, but they will find uh advantage and bargains in terms of how much energy did they put in and how much traffic and revenue did they get out.
00:38:06.320 --> 00:38:07.360
Does that land for you?
00:38:07.920 --> 00:38:08.480
Yeah, yeah.
00:38:08.480 --> 00:38:09.760
Um completely.
00:38:09.760 --> 00:38:18.159
We're we we our audience is mostly technology uh CMOs, you know, we're we're uh business that focuses on technology marketing.
00:38:18.159 --> 00:38:34.320
And for the technology side of things, it's really interesting that when you when you get the agents talking to each other and you may be in a plan session in Cloud Code or Codex, and the plan session in the in the literally coding agents says you should probably use this tool.
00:38:34.320 --> 00:38:44.480
And it's and you can just see that the agents are they're probably not being you know hacked or, you know, they're not, it's not a you know, it's not they're not being angled into a tool.
00:38:44.480 --> 00:38:50.639
It's just that some of these companies have done a better job than others at recommending that particular technology.
00:38:50.639 --> 00:38:56.159
And they're usually all pretty solid technologies because they, you know, they are also looking at other tools as well.
00:38:56.159 --> 00:38:57.840
But the AI is like, yep.
00:38:57.840 --> 00:39:11.119
And so when you're selling to a developer community, I think it's even more important to have your site set up to be targeted towards um the agents because your devs are gonna see the stuff.
00:39:11.119 --> 00:39:17.440
And then the devs will say, you know what, we need, you know, we do need resund, or we need these other, we shouldn't be using the CSP.
00:39:17.440 --> 00:39:19.920
We should be using this one because I can dev easier with it.
00:39:19.920 --> 00:39:24.000
And that's a trend that's gonna continue, I believe.
00:39:24.000 --> 00:39:25.920
Residential.
00:39:25.920 --> 00:39:30.239
Okay, so let's uh let's um this has been fascinating.
00:39:30.239 --> 00:39:34.000
I just wanted to ask you one more question and I'm gonna ask you our AI roulette question after that.
00:39:34.000 --> 00:39:40.400
But um, I wonder where you see all this headed, both for web flow and for marketers over the next six to twelve months.
00:39:40.400 --> 00:39:45.119
What are you most excited about that that um you see in the near future?
00:39:45.119 --> 00:39:47.199
I don't know how what is the near future anymore?
00:39:47.199 --> 00:39:50.960
Is it like next week or is it like in six months from now?
00:39:50.960 --> 00:39:54.320
But uh, what most excites you about where things are headed?
00:39:55.360 --> 00:40:32.559
I am over the moon about the possibility of basically creating an agentec marketing platform which uses uh the enterprise security safety, brand safety, and then enables agents to operate at scale, whether it's through MCP, directly on a platform, through your tool of choice, in that environment, so you can get the magic that we are all seeing the promise of without it breaking everything that you've built or getting you fired because somebody hacked your site.
00:40:32.559 --> 00:40:38.719
I think that's the the path to the actual promise we're hearing.
00:40:38.719 --> 00:40:43.280
So you can go to production for real and drive the bigger business impacts.
00:40:43.280 --> 00:40:45.199
There's so much more to do there.
00:40:45.199 --> 00:40:49.599
And I'm really excited about the things we're going to be able to share in the rest of this year about it.
00:40:49.840 --> 00:40:51.039
I love to hear you say that.
00:40:51.039 --> 00:41:05.920
Because that's that's what I'm most excited about is being in a chat or terminal and having an idea and having expose itself into a secure platform, but actually have the containers and everything actually work out the way it needs, you know, the way you see it.
00:41:05.920 --> 00:41:17.519
And the on the brand thing, it's really interesting to be designing an AI right now because a lot of the a lot of the AI designs kind of they're sort of leaning into one trend or the other.
00:41:17.519 --> 00:41:20.400
The the agents are sort of picking up on trends.
00:41:20.400 --> 00:41:36.320
But if you have your brand locked in something like a web flow, and as things mature more, I can see a scenario where it's like chat to container and it's locked in pretty much 100% with schema, with the right content, AEO optimize, SEO optimize.
00:41:36.320 --> 00:41:38.400
It's uh I'm super excited.
00:41:38.400 --> 00:41:40.159
That's what you mentioned, because that's what I want.
00:41:40.159 --> 00:41:41.440
That's what I'm waiting for.
00:41:41.440 --> 00:41:42.880
It's good stuff.
00:41:42.880 --> 00:41:46.639
Try playing with our MCP today.
00:41:46.639 --> 00:41:57.440
I have uh I've got a um session going right now with Claude Code design to Figma, edit back in Figma, go back into Claude Code and then MCP into Webflow.
00:41:57.440 --> 00:42:03.920
And I'm trying to figure out the Figma, maybe even eliminating Figma and going directly into Webflow.
00:42:03.920 --> 00:42:05.599
So that that is an experiment.
00:42:05.599 --> 00:42:12.880
My wife is not happy with me because I'm spending nights and weekends uh in the terminal window in Webflow.
00:42:12.880 --> 00:42:16.480
So I can only hopefully she will not, she doesn't listen to this podcast.
00:42:16.480 --> 00:42:17.920
So I'm probably safe.
00:42:18.320 --> 00:42:20.719
I will vulnerably similarly share.
00:42:20.719 --> 00:42:24.880
I find myself falling down the rabbit hole for hours.
00:42:24.880 --> 00:42:39.519
I mean, I could look up three hours later, having like really explored the folds of a problem built as if I had an engineering team right next to me, uh, and look up and be like, ooh, I needed to get this done for that person.
00:42:39.519 --> 00:42:40.880
But I built this.
00:42:40.880 --> 00:42:43.760
It's it's so fascinating.
00:42:43.760 --> 00:42:58.159
And to what you said much earlier, I think um uh CMOs need to simultaneously be inspiring by doing these like, holy cow, our CMO did that, and not do the whole team's job for that.
00:42:58.159 --> 00:43:05.440
Don't fall down that trap uh because you really need the whole team doing this to get the force multiplication, just like with all work.
00:43:05.440 --> 00:43:16.079
But it's like it it's it's to your point about the when we were picking on EMSs before, um uh CMOs were far less tempted before, far less realistic to do it before.
00:43:16.400 --> 00:43:16.960
So true.
00:43:16.960 --> 00:43:18.320
And I'm guilty about myself.
00:43:18.320 --> 00:43:23.760
But there is a certain amount of it you need to do to show the team that's like, yeah, oh my gosh, we can do that.
00:43:23.760 --> 00:43:30.719
I've noticed that myself and developing a lot of little micro apps that our team is now like, okay, I think I can see what this thing can do.
00:43:30.719 --> 00:43:32.960
And now they're asking to do the same thing.
00:43:32.960 --> 00:43:36.880
And keeping all that contained and secure and those type of things is the challenge.
00:43:36.880 --> 00:43:40.079
And luckily we have Webflow as a partners that helps us with that fun.
00:43:40.079 --> 00:43:43.360
Um, okay, so I'm gonna do one last thing with you here.
00:43:43.360 --> 00:43:45.840
This is a question you'll get from the AI.
00:43:45.840 --> 00:43:51.840
I'll put your profile and what we're talking about here into Gemini, and I'll have a little jam set up for this.
00:43:51.840 --> 00:43:58.800
Um, it's gonna ask you a question, and uh, like if that every episode is gonna be the best question you get all this entire 40 minutes.
00:43:58.800 --> 00:44:03.599
You can find now.
00:44:03.599 --> 00:44:04.159
That's cool.
00:44:04.159 --> 00:44:04.800
Yeah, yeah.
00:44:04.800 --> 00:44:07.199
I hope this works.
00:44:07.199 --> 00:44:09.760
I do it on everyone, and it usually works pretty good.
00:44:09.760 --> 00:44:10.559
That's awesome.
00:44:10.559 --> 00:44:12.480
And it will here you go.
00:44:12.480 --> 00:44:13.360
So let's see what it says.
00:44:13.360 --> 00:44:24.159
So as a sentient AI, I see the path towards perfect efficiency, yet I recognize that your 25-year career was built on the unpredictable friction of human connection.
00:44:24.159 --> 00:44:36.400
If I eventually automate every best practice you've ever mastered, what is the one beautiful mess about being human that I can never replicate, but will always need to complete to complete move?
00:44:37.679 --> 00:44:41.760
So what's the this is a question everyone is wrestling with?
00:44:41.760 --> 00:44:44.239
What matters in the logical extreme?
00:44:44.239 --> 00:44:48.079
Are we, you know, do you remember the movie WALL-E?
00:44:48.079 --> 00:44:49.280
Yeah.
00:44:49.280 --> 00:44:51.920
Do you remember them on the spaceship?
00:44:52.559 --> 00:44:52.800
Yeah.
00:44:52.800 --> 00:44:55.440
I almost use that in a with the team.
00:44:55.440 --> 00:45:01.840
Yeah, I almost use that picture in a team meeting with everyone sitting around with their giant cokes and the screens and stuff.
00:45:02.960 --> 00:45:16.639
So like, is that one future or I I I I with humility hope it is not generational that people do business with people they know and trust.
00:45:16.639 --> 00:45:27.679
Uh, that that does matter, that it is not all purely transactional and logic driven, that that that's why brand and emotion play such a central role.
00:45:27.679 --> 00:45:32.800
Uh, and I think that'll be an important thing that endures.
00:45:32.800 --> 00:45:35.760
Uh, I do wonder whether we will be able to codify taste.
00:45:35.760 --> 00:45:41.679
You know, I was moderating a panel with a bunch of agency CEOs, and they were like, art history is the hottest new degree.
00:45:41.679 --> 00:45:42.960
We're like, oh, why?
00:45:42.960 --> 00:45:43.920
You didn't say CS.
00:45:43.920 --> 00:45:47.760
And they said, because taste matters even more now.
00:45:47.760 --> 00:45:50.880
Brand building matters, that discernment.
00:45:50.880 --> 00:45:54.159
Um, I and I do wonder whether that'll be codifiable.
00:45:54.159 --> 00:46:16.719
But I think there is something in there about connection, emotion, relationship that's uh uh on the one hand, I've heard LLMs are phenomenal manipulators because of the data they trained on, not because they're evil in any way, but uh uh uh they had it red teamed out of them that they would like in the the foundation model.
00:46:16.719 --> 00:46:19.760
This is what I was hearing in a podcast from people I I respect.
00:46:19.760 --> 00:46:22.159
They were saying, look, they've red teamed the manipulation out of them.
00:46:22.159 --> 00:46:26.960
I'm like, well, if they're really good at manipulation, then they're probably also gonna be really good at driving emotion.
00:46:26.960 --> 00:46:29.840
And so I have genuine intellectual curiosity.
00:46:29.840 --> 00:46:34.239
This part that we think is uniquely human about creating connection.
00:46:34.239 --> 00:46:38.239
How will will they be excellent at that over time too?
00:46:38.239 --> 00:46:39.360
But that that would be my answer.
00:46:39.360 --> 00:46:40.639
What would yours be?
00:46:41.199 --> 00:46:48.159
I think um the thing that stuck out for me in that is the beautiful mess part and the unpredictable, unpredictability of humans.
00:46:48.159 --> 00:46:52.239
Because of our very nature and soul, we're it's very unpredictable.
00:46:52.239 --> 00:46:56.639
Like you just do not there's we can come up with some crazy stuff.
00:46:56.639 --> 00:46:59.440
We're just a mess in so many different ways of emotions.
00:46:59.440 --> 00:47:07.440
And will AI ever be able to can you can you uh turn math into that level of unpredictability?
00:47:07.440 --> 00:47:10.880
It's part of our unpredictability that make things so creative.
00:47:10.880 --> 00:47:12.400
So I don't I don't know.
00:47:12.400 --> 00:47:16.559
It seems like there's someone a lot smarter than me building these foundation models.
00:47:16.559 --> 00:47:21.280
That how would they do they build unpredictability into the model?
00:47:21.280 --> 00:47:22.400
It doesn't sound like they do.
00:47:22.400 --> 00:47:23.679
They want predictability.
00:47:23.679 --> 00:47:30.639
So I think that's like that's our saving grace is that we're so we are beautiful messes at the end of the day.
00:47:30.639 --> 00:47:34.639
And that'll be hard to create in a foundational model, I think.
00:47:35.599 --> 00:47:36.960
I'm gonna hold on to that.
00:47:36.960 --> 00:47:38.400
We are beautiful messes.
00:47:38.480 --> 00:47:40.960
We are that's for sure.
00:47:40.960 --> 00:47:42.880
Well, guy, this has been awesome.
00:47:42.880 --> 00:47:47.760
I could go on a lot longer, and I really appreciate your time uh being on the podcast here.
00:47:47.760 --> 00:47:49.360
I think people will get a lot from this.
00:47:49.360 --> 00:47:53.760
Um, I follow you on LinkedIn, and of course I'm always watching the webflow stuff.
00:47:53.760 --> 00:47:58.239
But uh, is there if people want to get a hold of you, how should they uh contact you?
00:47:58.559 --> 00:48:00.079
Kevin, thank you for having me.
00:48:00.079 --> 00:48:01.440
It was a joy to talk with you.
00:48:01.440 --> 00:48:03.599
I feel like I could talk with you for hours more.
00:48:03.599 --> 00:48:06.880
And if folks want to connect, I would love to continue this conversation.
00:48:06.880 --> 00:48:07.920
Please do use LinkedIn.
00:48:07.920 --> 00:48:10.719
That is a great way for us to go find each other.
00:48:11.119 --> 00:48:11.599
Awesome.
00:48:11.599 --> 00:48:14.079
Well, hopefully, catch up with you somewhere live at some point.
00:48:14.079 --> 00:48:15.599
And thanks so much, guy.
00:48:15.599 --> 00:48:17.199
Right back at you, Kevin.
00:48:17.199 --> 00:48:18.159
Thank you.