Cognition hires Art Levy as VP of Global Partnerships, targets one acquisition per week as it scales Devin worldwide

Speaker 1: Love it. Well, congratulations, and thank you so much for coming on. Thank you

Speaker 2: for your service.

Speaker 1: We'll talk to you soon. Thanks so much, guys. Yeah. Appreciate it. Goodbye. Up next, we have Art Levy, the new vice president of global partnerships with Cognition. We got Russell Kaplan, the president of Cognition. They're coming in to the TBPN Ultra Dawn. Very excited to have some forks joining. Congratulations.

Speaker 2: There they are.

Speaker 1: The trade

Speaker 2: In the corner office. In the corner office. For a new vice president.

Speaker 1: Corner office. Yeah. Where are you guys in the South or Bay Area?

Speaker 8: We're calling from Austin. Yeah. Calling from our Austin, Texas office.

Speaker 1: Austin, Texas.

Speaker 8: We got a we a great team out here. Cool. Yeah. Super excited to be having Art on the team.

Speaker 1: Okay. What is a global partnership? Does it have to be global? Is it just a sales deal, or is there a deeper partnership that happens with these companies like AWS, Apple, Slack, Nivon, Oracle, Capital One? Blur break it down.

Speaker 11: John, you nailed it. Global partnership means it's not just a sell to deal.

Speaker 1: It's Okay.

Speaker 11: Something where we're both selling to the business. Potentially, there's an investment. Potentially, we're co selling to their customers. We're selling to them. We're selling to their customers. We're doing co marketing. We're doing a a press release. Potentially, there's a product integration. It's about deeply connecting to businesses who share common goal around, you know, acquiring same customers or we build inside of them a product that is just better for their customers, and then both parties win. You know, classic partnerships, one plus one is three.

Speaker 1: I think of cognition as an AI lab, maybe a neo lab, a harness developer, a consulting firm in some world.

Speaker 2: I think of it as a source of geopolitical power. Guess. That's just me personally.

Speaker 1: Yes. It's it's birthing a new god as well. But but what what is the conversation that you're having with Fortune 100, these huge companies? How are you positioning your role in the AI ecosystem, in the business ecosystem? How what what what is this full scope of the value that you can deliver

Speaker 8: to customers? As as we've as we've grown up and evolved, you know, we've I think we've really realized our our role is to be the independent agent lab. Yeah. And so we're an applied AI lab. We're focused on building agents. And we want to give every company the best possible agents to go solve all of their software engineering problems. And so whether it's a Fortune 500 or it's a small, fast growing startup, people have a lot of needs for software. And Devon, from the beginning, has always been designed to work with all the other tools you already use. Right? It shouldn't be, you know, it's the Devon way or the highway. Like, Devon should join you like a teammate. And I think that's one of reasons I'm super excited that Art is joining us. Art built an amazing alliance and partner ecosystem at Brex over his, like, very long career there from employee thirty's chief business officer. And we see an opportunity in competition to bring not only, you know, it's a it's not just a, hey. Here's the tool. Go figure out. It's like, how does this work super well with every piece of technology you have inside your organization?

Speaker 1: Are large corporations that are maybe they don't have an in house AI lab, are they generally of the view AGI is real, AGI is soon, but we're going to go through a sort of slow takeoff, diffusion adoption will be a big problem, and cognition can help sort of realize that thesis?

Speaker 8: I think the pace of progress in artificial intelligence is dramatically over the pace of progress of internal organizational redesign, for By the way, this is true not just at the big Fortune 500s. Even the sort of startups or tech companies we would consider as fast movers, everyone is asking, do I really need to rethink everything now because of AI? And I would say it's actually one of the things that's been lucky for us at Cognitions. We only incorporated it in January 2024. So our internal team setup and structure, it's very AI native. But every company is trying to figure this out for themselves. And I think if you froze capabilities today, you have at least a decade of crazy progress in the full enterprise world just to get this stuff really rolled out and operationalized.

Speaker 1: So what are like, when you're talking to an enterprise, is is it better to position as the the, you know, you're gonna help with this diffusion, this value creation? Or are we still in that sort of, I maybe it was, like, May, June, July, like, woah. We went a little bit too far with the token maxing. Maybe Cognition can help me rein in this budget, actually make sure that we're getting ROI. What's resonating there on the on the on the cost of benefit position?

Speaker 8: I think I think it's both. Think it's the third tier. So first, you know, all the models have been getting better. Right? Yeah. And so as the models get better, the capabilities keep improving. Mhmm. Yeah. People are starting to ask the question, hang on. Like, do I need to drive the Ferrari to the grocery store? You know, can I can I use an open source model for parts of these workloads? Can I use cheaper model, faster inference? Yep. And so that's definitely one of driving

Speaker 2: a Ferrari to the grocery store.

Speaker 1: You're like, I don't believe

Speaker 2: But continue.

Speaker 8: If only we could all be Jordy.

Speaker 2: All your

Speaker 8: help, Jordy. But guess one part of it. I also think there's a separate thing that's emerging, which is if you actually want to have true frontier capabilities, different models have different spikes. Right? Like, we've seen, for example, we've seen the the OpenAI models, for example, are are the best in the world right now for, for recall of security vulnerabilities. Like, if you wanna catch if you wanna catch as many as possible, the OpenAI model series is best. There might be others where actually the anthropic ones are a little bit better on precision, getting it really, really tight in. So I think there's, in addition to the price performance angle, for enterprises, there's this element of, well, do I use the best tool for the job, for each job to be done?

Speaker 1: Yes.

Speaker 8: And then there's the people change side, which we try to help with too. I'll give you one story on this. We had a customer, global 2,000 customer, tens of thousands of engineers, and they deployed Devon across the entire organization. And they were seeing crazy productivity gains. But one organization they deployed to was the product management organization. And so every product manager now had Devon. And what they were doing was that this organization's job is basically taking customer requests and turn them into Jira tickets, roughly. It's like scope out what the customers want in of Jira tickets. So they were using Devin to they took those requests, and then they had Devin actually go prototype to fix. And then once Devin prototyped the fix, they said, Devin, can you summarize this fix in a Jira ticket? And then they deleted all the code. And we asked them, guys, what are you doing? You already had fix. And the product manager we talked to said, that's not my job. I'm part of the product manager organization. I'm not part of the engineering organization. And so I think for really large companies who've been doing things the same way for a long time, there is a total rethink that needs to happen in just literally how you design your structure.

Speaker 1: Yeah.

Speaker 2: How are you guys thinking about M and A? You bought Poke I think it was surprising, but made a lot of sense. It was surprising to me because I figured someone like an Apple or something would would see Poke's, you know, talent and their kind of consumer insights and all that kind of thing, try to pick them up. But how are you guys thinking about m and a from like where the value really comes from? You guys are very good at building software. That's your whole pitch. Right? Is where you help you build software. So I can imagine a world where product acquisitions are like less interesting because you're looking at your own road map and you're looking at what someone's built and thinking like, well, how quickly can we build this ourselves? Is it is are we should we expect, like, more sort of, like, talent style acquisitions? What where where what's your general framework, and and maybe what's share as much of your mandate as you can, Art.

Speaker 8: Yeah. I mean, would say all of the above. I mean, we have today in Cognition, 97 former founders working in the company to an after all startup. And we are we are, like, aspiring for Apple hiring. One one team a week probably right now. Oh. So if you're if you're a Trump value team and you want to be part of rocket ship, get in touch. Love bringing on founders. But we also think there's bigger companies out there with products, with distribution, with learned insights that are really useful for us that we want to bring in as well. I think how you do that successfully, a lot of companies screw up M and A. We kind of learned learned it trial by fire acquiring Windsurf in the course of a weekend as a company that was four times our size from, like, first call to executed reading from Friday night to Monday morning. So it's something that I think we're only going get better at. And one reason I'm really excited artists here is to help us level up that process and be the best destination in the world to be acquired by.

Speaker 11: Yeah. I think one of the things I've seen work well in my career is a lot of times partnerships can be a a preempt in m and a. So a bit of a try before you buy. You make sure that the cultures really make sense to do companies build an integration or a partnership. And suddenly in those conversations or after you delight a few customers, you both decide, hey. Actually, maybe it'd be better if we work together. And I think given the scale of cognition or rate of growth, the impact that founders can have when they come here is really gonna be generational. So I'm really excited to, you know, talk to a bunch of founders and hopefully bring a bunch more of them onto the Cognition

Speaker 2: team. Awesome.

Speaker 1: What country internationally outside of The United States is punching way above its weight in terms of AI adoption, AI diffusion? You always hear that story about Estonia rolling out high speed bandwidth for the entire country, and they just went way forward in terms of GDP per capita. What country are you having, like, surprising success in?

Speaker 10: Japan. Japan.

Speaker 8: Yeah. I would say Japan, number one. We love Japan. Look, we're a global company now, so we have people all around the world. We have customers all around the world. But I a got a shout out Japan in particular. Even from the beginning of when Devon first became available self serve, cloud agents were just at the edge of possible, we had an enormous uptake of users in the Japanese market. And I think there's a few reasons. I think one of them is our Japanese customer base, everyone is incredibly conscientious and detail oriented. And whenever we get a support ticket from Japan, it's a very thoughtful essay of every single issue that has been tried, every piece of debugging information you would need. And like they just found a bug. They they just found a bug. They if they could have solved it on their own, they wouldn't. And so I think the user base in Japan is super sophisticated, really early adopters, and and it's been exciting to see all the growth there.

Speaker 1: I prefer the American style of just fix it. Fix it. No. It seems it seems like really beneficial. Sorry. Are you you were gonna say something?

Speaker 11: No. Was gonna say, as I'm digging in, also ironically on the channel sales side, we've actually had the most success with sell through with partners in Japan who seem to just deeply understand the product and the value prop and be able to sell it to to their customer bases.

Speaker 2: Interesting. Do do the Japanese love otters by any chance? Could that be a factor?

Speaker 8: That's a good that's a good question. We we definitely

Speaker 2: see the plushie.

Speaker 8: Yeah. Yeah. Got a lot of we got a lot of otters fighting with There was a big debate in Japan of the right honorific for Devin. You know, is it is it Devin Son, Devin Koon, Devin and then we we learned that the community in Japan has settled on Devin Koon as the right honorific, you know, your your friend, your buddy, your your sidekick who's who's helping you out. I think it's a it's a fitting name for

Speaker 1: a And that came from the community. This was not internal?

Speaker 8: Came from the community. Yeah. We were informed. We were informed. Ground out.

Speaker 1: Ground out, man. Amazing. Okay. Well, thanks for breaking down. That's very

Speaker 2: Love it. Well, I'm excited to see everything that you guys do together.

Speaker 1: Yep. Deals. Deals. Deals. Very excited. Awesome.

Speaker 8: Great to

Speaker 2: see Talk

Speaker 1: to soon.

Speaker 2: Great hanging, guys. I'll see you there. Congrats.

Speaker 1: Goodbye. Let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents, whether you're writing code, analyzing data, creating content, or automating business workflows. Codex helps you move projects forward from start to finish. We have Brendan Carr from the FCC joining the show shortly. It's his first appearance, but we met years ago. I've always been fascinated by his perspective on everything that the FCC does. There's obviously expanding five g. We can talk about satellites and space Internet, what it takes to launch a rocket. I know you've been thinking about the TBPN satellite cluster. He's the guy to talk to.

Speaker 2: Perfect. TBJ over on the X Chat says literal art of the deal right here. Wow. Insane Who? Nominative determinism.

Speaker 1: Art of the deal. While we are waiting for chairman Brendan Carr to join, we gotta go through this crazy historic analysis from Ed Zitrin.

Speaker 2: Ed Zitrin, 07/29/2024 gave Sam Altman the perfect road map. He said, I am hypothesizing that for OpenAI to survive for longer than two years, it will have to. And then he laid out pretty much