# One Founder Now Runs a Game Studio With 20 AI Bots | Walden Yan

Markdown copy of https://realalpha.app/v/fFS4bs2Q9_c/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Walden Yan says Devin Bots can turn agents into proactive coworkers and shift engineers from reviewing code to setting goals and guardrails.

- Show: MTS
- Hosts: Madison Kanna
- Guest: [Walden Yan](https://realalpha.app/builders/walden-yan/) is co-founder of Cognition.
- Published: Nov 18, 2026
- Length: 32 min

## The best minutes

- [17:14](https://realalpha.app/v/fFS4bs2Q9_c/?t=1034) [Agents can take ownership of measurable organizational goals](https://realalpha.app/s/agents-can-take-ownership-of-measurable-organizational-goals/) (3 min)
- [24:33](https://realalpha.app/v/fFS4bs2Q9_c/?t=1473) [Code review can fade as teams build stronger guardrails](https://realalpha.app/s/code-review-can-fade-as-teams-build-stronger-guardrails/) (3 min)

## Also worth watching

- [0:00](https://realalpha.app/v/fFS4bs2Q9_c/?t=0) [One founder's 20 bots raised output to 1,900 PRs a week](https://realalpha.app/s/one-founders-20-bots-raised-output-to-1-900-prs-a-week/) (3 min)
- [3:00](https://realalpha.app/v/fFS4bs2Q9_c/?t=180) [Poke's proactive assistant design became engineering bots](https://realalpha.app/s/pokes-proactive-assistant-design-became-engineering-bots/) (3 min)
- [12:22](https://realalpha.app/v/fFS4bs2Q9_c/?t=742) [Project context lets bots surface related work proactively](https://realalpha.app/s/project-context-lets-bots-surface-related-work-proactively/) (2 min)
- [20:26](https://realalpha.app/v/fFS4bs2Q9_c/?t=1226) [Agents may break organization-scale problems into experiments](https://realalpha.app/s/agents-may-break-organization-scale-problems-into/) (2 min)

Skip about four minutes of sponsor reads, the custom-software discussion, and the ambition discussion and closing pitch.

## Full analysis: Walden Yan says weekly PRs rose from around 50 to 1,900 with 20 bots

- Yan said a solo founder with 20 bots went from around 50 to 1,900 PRs weekly.
- That is a striking throughput report, not evidence bots caused higher revenue or ratings.
- Track accepted work, quality, and spend before treating one studio as a staffing plan.

Its 6 sections, what it means for you and what to watch are on [the page](https://realalpha.app/v/fFS4bs2Q9_c/#analysis), free with an account:

- [0:21](https://realalpha.app/v/fFS4bs2Q9_c/?t=21) One founder's bot team multiplied reported PR throughput
- [4:26](https://realalpha.app/v/fFS4bs2Q9_c/?t=266) Cognition says bots already build Cognition's products
- [17:48](https://realalpha.app/v/fFS4bs2Q9_c/?t=1068) Outcome goals can make agent spending governable
- [12:39](https://realalpha.app/v/fFS4bs2Q9_c/?t=759) Shared context gives bots room to act proactively
- [24:49](https://realalpha.app/v/fFS4bs2Q9_c/?t=1489) Less review requires stronger system-level controls
- [21:54](https://realalpha.app/v/fFS4bs2Q9_c/?t=1314) Yan expects agents to plan organization-scale experiments

## Summary

Walden Yan is a co-founder of Cognition. He discusses Devin Bots and says Cognition uses Devin to build its products, including its Android app.

Yan says Devin Bots are designed to take on ongoing roles rather than isolated tasks, and describes a one-person game studio using 20 specialized bots that went from around 50 to 1,900 PRs a week. He says Cognition uses Devin to build its own products, including an Android app maintained by a bot. Yan argues that people should manage agents through decisions, goals and guardrails rather than oversee every task or line of code. He expects code review to recede as engineers build systems that let agents work more autonomously.

### Minute by minute

- [1:35](https://realalpha.app/v/fFS4bs2Q9_c/?t=95) **Bots take on ongoing roles.** Yan says Devin Bots take on jobs, such as owning a release process, rather than single tasks. He describes a game studio run by one founder and 20 specialized bots: its output rose from somewhere around 50 PRs a week to 1,900, and Yan says the game’s ratings and revenue also increased.
- [3:00](https://realalpha.app/v/fFS4bs2Q9_c/?t=180) **Poke inspired proactive agents.** Yan says Cognition acquired Poke, whose iMessage agent reminded users about things they had forgotten and helped them keep up with work. He says Cognition saw a chance to bring Poke’s personal, proactive and persistent agent technology to engineering teams.
- [4:05](https://realalpha.app/v/fFS4bs2Q9_c/?t=245) **Cognition uses Devin to build Devin.** Yan says Cognition has long used Devin to build Devin and that he struggles to find code not written autonomously by it. For the Android app, the bot was told to maintain parity with the iOS app; Yan says it can go days without input while merging thousands of PRs and testing its work.
- [5:43](https://realalpha.app/v/fFS4bs2Q9_c/?t=343) **Decisions reduce agent oversight.** Yan says the decisions feature helps people managing parallel work focus on issues that need their attention, while low-risk changes can be merged automatically. He says a bot can also start and manage a team of agents for a larger job, reducing the overhead of coordinating them.
- [10:09](https://realalpha.app/v/fFS4bs2Q9_c/?t=609) **Agents can maintain custom tools.** Yan argues that people can build and maintain professional tools tailored to their work, rather than treating AI-built software as a toy. He says an agent can also handle scheduled, intelligent tasks itself, so a user may not need to build a separate app for every need.
- [12:22](https://realalpha.app/v/fFS4bs2Q9_c/?t=742) **Proactivity depends on context.** Yan says proactive agents need to build context about a person over time, using sources such as GitHub and company Slack. He says bots have flagged colleagues working on related projects and prepared changes that address issues affecting a user’s work.
- [16:56](https://realalpha.app/v/fFS4bs2Q9_c/?t=1016) **Teams can set outcome-based goals.** Yan says engineers need to move from thinking like individual contributors to managing agents. Instead of assigning tickets, teams can give bots goals such as improving application speed or onboarding retention; he argues that tying outcomes to budgets can help enterprises manage AI spending.
- [20:36](https://realalpha.app/v/fFS4bs2Q9_c/?t=1236) **Agents may take on broader problems.** Yan says the next step is moving from individual engineering tasks to goals for an entire organization. He says bots can choose models for tasks based on price and performance, and hopes that within a year they can break down problems such as drug discovery into experiments, software needs and investment plans for people to evaluate.
- [22:35](https://realalpha.app/v/fFS4bs2Q9_c/?t=1355) **Guardrails could reduce code review.** Yan says code review may soon become unnecessary as engineers focus on architecture and guardrails instead of reading every line. He points to security-conscious code structure, formal verification and better testing; for example, prompt changes at Cognition must be tested at least 10 times, while some changes still need human review.
- [27:17](https://realalpha.app/v/fFS4bs2Q9_c/?t=1637) **Ambition can unlock more uses.** Yan says he was struck by a founder running a game studio with 20 bots and urges users to try more ambitious projects. He says Devin Bots fulfill his original goal of having an agent maintain 10 or 20 side projects, which he says is now possible with improved cloud infrastructure and multi-agent capabilities.

## Transcript

MTS, hosted by Madison Kanna, with Walden Yan.

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[0:00](https://realalpha.app/v/fFS4bs2Q9_c/?t=0) One of our early beta testers, she runs a gaming company. The gaming company is just her and 20 different specialized bots of different roles. And those bots do everything from the UI design to the release management. They have these various jobs. They coordinate with each other. They do daily stand-ups. They give each other's kudos.

[0:21](https://realalpha.app/v/fFS4bs2Q9_c/?t=21) **Walden Yan:** And her PR throughput, she was shipping, I think, somewhere around 50 PRs a week at the start of it. She's shipping 1,900 PRs a week by the end of it.

[0:30](https://realalpha.app/v/fFS4bs2Q9_c/?t=30) And it's actually an award-winning game. And, you know, you can see that the ratings of her game and revenue has actually increased over the period that she's done this. Hey, welcome back to MTS. I'm here with one of the co-founders of Cognition, Walden. Thanks so much for joining me, Walden. Yeah, no, I'm super excited to be here and excited to be chatting with you. I know we gave you access to DevonBots a little bit earlier. I hope that's been fun to play with. Yeah, they've been fantastic to play with. I think the timing was great because I was playing with the Devonbots a couple of days ago before the release. I got early access. Thank you. And then I had been talking to you guys and you didn't know that I was about to join MTS. And so when I heard that you were going to do this interview, I had to say, I want to do the interview because I've been playing with the bots. And so it's really exciting to talk to you.

[1:19](https://realalpha.app/v/fFS4bs2Q9_c/?t=79) No, I appreciate you beta testing it. No, we've been getting a lot of beta testers onto DevonBots over the last couple of weeks. And especially on engineering teams, the reception has been incredible. And so we appreciate everyone who's been trying it out. Yeah, amazing. Okay, so for someone who is listening and they're just hearing about DevonBots today, they just launched, congratulations. Can you tell me about them for someone just who doesn't have any context yet? Yes. Hello, everyone online. The way I think about DevonBots, normal agents, you give them individual tasks like, hey, go do this for me.

[1:54](https://realalpha.app/v/fFS4bs2Q9_c/?t=114) Oh, go implement this PR, change the color of this button. DevonBots are built to be much more durable and proactive coworkers. And so you give them a job, really, rather than a task. So, hey, I want you to just own our release process. Or, hey, I want you to be my chief of staff, for instance. And really, it's very engineering focused. So one great example I've seen of this, one of early beta testers, she runs a gaming company. The gaming company is just her and 20 different specialized bots of different roles. And those bots do everything from the UI design to the release management. They have these various jobs.

[2:38](https://realalpha.app/v/fFS4bs2Q9_c/?t=158) They coordinate with each other. They do daily stand-ups. They give each other's kudos. And her PR throughput, she was shipping, I think, somewhere around 50 PRs a week at the start of it. She's shipping 1,900 PRs a week by the end of it. And it's actually an award-winning game. And, you know, you can see that the ratings of her game and revenue has actually increased over the period that she's done this. So we've seen people use DevonBots really as more, like, true autonomous coworkers. Yeah, that's really cool. OK, I want to take it back a little bit and ask, like, what was more of the inspiration behind building them when you were starting out?

[3:12](https://realalpha.app/v/fFS4bs2Q9_c/?t=192) Yeah, absolutely. You may know a few months back, we acquired a company, Polk. Polk, they were really the pioneers of what were actually personal, proactive and persistent agents. And so the design of Polk was that it's an iMessage agent and you could text it and it would remind you of things that you forgot about. It would help you stay on top of your work. Honestly, it was a hit. People really loved it. I think that we saw an opportunity when we acquired them to take that same technology and apply it to engineering teams. Because really, it's not just a personal assistant that people want to use AI for.

[3:49](https://realalpha.app/v/fFS4bs2Q9_c/?t=229) People really want to use AI to go accelerate everything that they're doing in their companies. People out there want to build more software. People want to enable completely new powerful things. we've highlighted in our videos. And so that technology, that acquisition and poke game have been a huge part into making this thing happen. Yeah, absolutely. So I want to talk about something else I saw. Someone from Cognition was saying that in many ways, the bots built themselves. How much of this is true? Can you tell me more about this process? This is really fascinating to me, the product building itself, essentially. Absolutely. So we've always used Devon to build Devon. I have a very hard time finding any code that's not written by Devon fully autonomously. In fact, many people at Cognition, we've moved away from laptops. Some people have moved to using iPads because you can actually just use the Devon app on your iPad, and it's fully in the cloud, and it's autonomous. That's only been possible because our entire development process has been moved to Devon. That's always been the case. Now, with Devon bots, the thing that has become insane is if you look at the number of PRs people merge compared

[4:58](https://realalpha.app/v/fFS4bs2Q9_c/?t=298) to the number of messages people send to their bots, that has gone up about 4x since we've used bots internally. So that means that every command or message you send now goes further. And so I'll give you a great example of this. Building our Android application, we've been building our

[5:13](https://realalpha.app/v/fFS4bs2Q9_c/?t=313) **Walden Yan:** mobile services for bots. Usually you would hire a team to go work on Android. Instead, we said, hey, look at our iOS application.

[5:21](https://realalpha.app/v/fFS4bs2Q9_c/?t=321) Your bot's job is to just fully maintain parity with the iOS application. And we gave it that one instruction. We'll check in only every now and then. There'll be days where we never talk to it. And it's merging thousands of PRs and keeping up to parity. And it's not slop. It's actually running the code, testing it, making sure that the pixel level comparisons are correct. This is kind of where we've really seen the throughput grow immensely with bots. Yeah. Yeah. And one feature that I liked when I've been using bots is the decisions feature. So basically it gets so overwhelming, like all of the different, you know, things you have to keep track of, like when is your agent done? Can you talk a little bit more about the decisions feature?

[6:04](https://realalpha.app/v/fFS4bs2Q9_c/?t=364) Yeah, no. Thanks for noting that. One of the things that people who are doing so many things in parallel notice is that there's a lot of context switching. There are a lot of things going on, and they really just want to know about what are the most important things. And so we have people where, yeah, you don't have to pay too much attention to the details of the conversation. You just know that, hey, if Devin bot actually needs me for something, it will give me an item to make a decision on. And so, yeah, I'm not going in, looking at the threads and reviewing every line of code. If Devin thinks something is important enough for me to review, I will talk about it high level at that decision point.

[6:41](https://realalpha.app/v/fFS4bs2Q9_c/?t=401) And then for things that are low risk, I will just go merge automatically. And so it's the right level of interface for what actually needs me to look at versus what the agent can handle on its own. Yeah, as I was using it, I almost couldn't believe there was something, I couldn't believe there wasn't something like this before, even six months ago, because there's just so many times I'm lost in my threads. I don't know what to go back to, right? And just constantly forgetting when you're trying to use these agents. And so I really like that it's kind of telling me like, what do I need to look at right now versus I'm trying to keep track of everything in my head.

[7:13](https://realalpha.app/v/fFS4bs2Q9_c/?t=433) Yeah, well, a year ago, if you look at people using Devon, And very few people would be trying to manage multiple threads in parallel. Come to today, many people are not only managing multiple threads in parallel, they're also trying to use devins to manage other devins. And so we realized that there's already this demand that didn't exist before for a better system to coordinate all these teams. And so this brings us to the other thing that really makes DevonBots different, I think, which is if you give it a goal or a job, the DevonBot itself will go spin up that whole team of agents and help manage it for you and basically reduce that overhead burden for yourself.

[7:52](https://realalpha.app/v/fFS4bs2Q9_c/?t=472) Yeah, I think that's the part I'm still getting used to. Basically, I'm used to managing individual sessions, right? but then at this this abstraction level, like you have a bot managing other bots. I'm still getting used to it. What is your, what are your tips or heuristics for like working in this new way? I think that many people I've talked to when I first showed them bots, they are not ambitious enough because they're, they are so used to being bottlenecked by what the agent can do for them. I would really say, take your most ambitious ideas and you can keep throwing them at your devon bought because it can work on multiple of these threads and parallel and in fact it's it's better because over time it compounds and it learns uh from what things you want what things you don't want um when we were filming with venus williams actually uh there's actually a lot of complexity that goes into being a professional tennis player i learned that you know it matters what altitude you're playing at and humidity and you have to string your rackets to the right temperature or to the right uh pressure to match that uh the the team that they work with has to get the exact uh specification of balls that match the tournament balls ahead of time uh there's

[9:04](https://realalpha.app/v/fFS4bs2Q9_c/?t=544) certain meal schedules and prep schedules you have to follow so there's actually a lot of things that go on behind her um and i think she wasn't used to before devon bots having something that could handle just every request that keep coming up but then i i really encourage her just call devon bots and just say what's on your mind. Say what else you need. And I think that was so powerful that you could just fully unleash your ambition. So yeah, my overall tip is don't hold back all the various things you want it to be able to do. You should just throw it all at it because it can handle it. I want to shout out our sponsor, Lovable.

[9:37](https://realalpha.app/v/fFS4bs2Q9_c/?t=577) You know the app you've been meaning to build? The internal tool, side project, or product you'd otherwise lose a weekend to? With Lovable, you get software you can ship today. That includes an editable code base you can inspect and change, plus two-way GitHub sync. Lovable also handles the backend and infrastructure, including managed Postgres, auth, storage, hosting, payments, and more, so you don't have to wire it all up yourself. And through Lovable's MCP server, you can create and deploy your project directly from the agents you already use. Turn ideas into software people love with Lovable.

[10:09](https://realalpha.app/v/fFS4bs2Q9_c/?t=609) Lovable.dev. Now, on to the episode. I'd love to talk more about this because this is actually something I noticed when I was first using my Devon bots um so I ran into this problem where you know I worked as an engineer for a decade and I think I'm used to doing things in an old school way like for example um I got into Devon I'm like okay build this app for me and then I realized I don't need to build an app like my bot Ada could just do this for me um and and then it was like from there I was thinking there are all these things where I'm thinking oh I need a whole app for this but really it's just my bot could do it entirely and run it, run it, run everything for me. Um, so I was thinking, yeah, if you could expand on that at all, like, how can I be even more ambitious with my bots? How can I be thinking more on like, how can I push the limits here more?

[10:59](https://realalpha.app/v/fFS4bs2Q9_c/?t=659) The thing that I, let me actually walk this, this back a little bit. We filmed our launch video with a lot of people who are engineers, but a lot of people who weren't engineers. And I think some people initially may be surprised by that. We think that we're entering a world where it's not just non-engineers get to vibe code stuff. I think we're past vibe coding now. I think we're at a point where people can actually build and maintain their own professional tools custom to what they want to do. And it's not some toy thing. It's actually professional tools for the thing that they want to spend all of their time doing. When you get a Devon bot to work on it, you have this interesting mix where you get to consistently customize what you are doing, but the agent itself is also continually being intelligent and can help you fill in the gaps. So if you want another feature or idea or to do something, the agent can always fill that in for you and it can always keep on being proactive and staying on top.

[12:00](https://realalpha.app/v/fFS4bs2Q9_c/?t=720) that for you. So I think your example was you were initially like, oh, I wish my app could message me on some schedule and be intelligent. It was like, actually, the agent can message you on the schedule and be intelligent. And it can still build apps and add that backing for you. But you never feel like you are limited by the software it's building. There's always that powerful thing behind it. Yeah, absolutely. And something else I liked about the bots is they felt like they were, you know, kind of preemptively like anticipating my needs. Can you talk like more about that and how they do this? I know it's a big question. Yeah, absolutely. The key thing about proactivity is really understanding you and getting context. So I think that was, that is job number one. And this is what the POKE team really brought on board when they came to work with us was understanding how do you build that context of this person over time, get a sense of what they might want.

[12:58](https://realalpha.app/v/fFS4bs2Q9_c/?t=778) How do you build the right connections to your GitHub if you're trying to understand the project they're working on or the company Slack if you're trying to understand what projects are coming up. That's where productivity can be very useful. So a great example of this, oftentimes internally when people are using Devonbots, the Devonbots will sometimes tell you, hey, did you know that there's another person over there working on something that seems very similar, very related to what you're doing? You should maybe go talk to them about that. Or, oh, there's this other change that actually does impact what you're doing.

[13:27](https://realalpha.app/v/fFS4bs2Q9_c/?t=807) I already put up a change for you to merge that will help fix that. That really creates this magical moment. And I think a lot of our focus continues to be how do we build something that feels more proactive, produce more of these magic moments for you. So the idea is you're kind of getting updates on what your coworkers were doing and how it's affecting your code base or whatever you're working on. Totally. And that's just one source of things that might come up, right? Like I could totally imagine that we'll have dev and bots that will notice, oh, like, did you see this thing in the news? Maybe like we could use this new technology in your product or, hey, like I just learned about, you know, for example, if you're studying methods for remembering things, I know that's the app that you're building.

[14:13](https://realalpha.app/v/fFS4bs2Q9_c/?t=853) Hey, like there's some new studies on new methodologies for this. Do you want to learn that? I can see a world where bots can really bring all of this into your life. Yeah, I think that was the part I was most excited about using my Devon bot. Ada was thinking, like, how can she maybe connect some of the things that I decided she was as she is, but how can she connect some of the things that I'm learning about? Like, for example, I learned about two different subjects and how can she maybe start linking these things together? That's something I'm excited to kind of push forward and see, like, where we can go from there.

[14:46](https://realalpha.app/v/fFS4bs2Q9_c/?t=886) Yeah, absolutely. Now that you've used Dev and Bots, I guess, for a little bit here, I'm curious, what are like kind of the biggest questions on your mind or maybe like some like, you know, what are the hardest hitting questions you have? We'll continue monitoring right after this message from our sponsors. ElevenLabs, AI that communicates at human level across every channel and modality. ElevenLabs.io slash MTS. Scale your startup on Neon. Millions of developers and startups have already chosen Neon for their backend. Start on the free plan or get up to $100,000 in credits for your startup at neon.com slash MTS.

[15:24](https://realalpha.app/v/fFS4bs2Q9_c/?t=924) Special thanks to our sponsor Kong, the AI connectivity platform. Connect APIs, LLMs, agents, and systems with serious security and governance. konghq.com Finally prove what your AI investment is worth with Lairdyn. Head to lairdyn.com today to start measuring your AI's bottom line. MongoDB, trusted to run critical applications by 75% of the Fortune 100, the intelligent data platform for the era. Apply today for access to up to $15,000 in Atlas credits, dedicated technical support, and exclusive events and go-to-market opportunities. MongoDB.com. Chargebee, billing and monetization infrastructure for the AI era.

[16:04](https://realalpha.app/v/fFS4bs2Q9_c/?t=964) chargebee.com ClickHouse is the real-time database built to keep up with every query your apps and agents send its way. And it now runs Postgres, observability, and AI tooling on one platform. Learn more at clickhouse.com You should fast. So do attackers. Don't assume you're secure. Prove it. Horizon 3. Go hack yourself with AI. visit horizon3.ai slash node zero the key ingredient in today's m&a srs aquium learn why their shareholder rep payments and escrow solutions have been used by 96 percent of top global pe firms and 88 percent of top global vcs at srs aquium.com slash mts that's s-r-s-a-c-q-u-i-o-m dot com slash mts i think it goes back to what i said earlier just like um i'm so used to all these individual sessions and uh one like the decisions future like getting used to that um but then also probably just like yeah a bot managing bots like i haven't done as much of that and so like shifting my mindset for like working in that way and and like figuring out how to do that exactly so far yeah you know there's always this learning curve for people using devon which was you kind of had to bring yourself from thinking about your job as an ic to almost feel like a manager of agents i think that's a little bit related to the shift of what

[17:36](https://realalpha.app/v/fFS4bs2Q9_c/?t=1056) you're thinking about which is like hey like there's kind of these new paradigms one thing that i think devon bots will actually really change for the industry is the way you interface with them is now another level of abstraction up, where organizations, instead of assigning tasks like, oh, here's your ticket, here's that ticket, here's that ticket, and giving tickets to their agents, they can give much more high-level but also much more impactful goals, like go optimize the speed of our application, go increase our user retention on our onboarding

[18:12](https://realalpha.app/v/fFS4bs2Q9_c/?t=1092) **Walden Yan:** flows. You can actually give very clear and quantifiable metrics and goals to your bots and trust that they will go now spin up a team to go optimize that for you. And by the way,

[18:23](https://realalpha.app/v/fFS4bs2Q9_c/?t=1103) when you look at companies like Uber who want to cut back on spend and are worried about their AI spend exploding, I think the only way you can really get to a point where you can be comfortable with AI spend is if you can give your agents these clear outcomes and definable goals. So this was the other big thing that inspired us to build DevonBots, was building something that can actually be outcome-oriented from first principles and allowing people to tie those outcomes to budgets. This is going to be, I think, a really big change for the way large enterprises in the world are thinking about deploying agents. Yeah, absolutely. And I know you've talked before, you've talked about enterprises where there's still so many companies out there that don't really actually know how to get started with AI, how to onboard, like how do DevonBots fit into like helping them getting started? In many ways, DevonBots, because it's proactive, will help you discover uses of your AI that you didn't even realize. And so one thing, just very clear example, is if you ask your DevonBot, hey, can you help me look at my analytics for a second? It might also automatically say, hey, do you want me to also send you daily reports on

[19:33](https://realalpha.app/v/fFS4bs2Q9_c/?t=1173) these things? Or, hey, do you want me to improve your data pipeline? Because they're proactive and you feel like you're working with a human, it can teach you new ways to use the tool itself. And so new users, I think, often get to this point where, like I said, they're not ambitious enough with their AI or they're not talking to it correctly. How do you hold this new tool? I think DevonBots can feel more like just a very proactive coworker where like, hey, I'm here to help you and here's all the way. I can help you that you may not have even known about. Yeah, absolutely. And I think that's like you asked me my biggest challenge.

[20:06](https://realalpha.app/v/fFS4bs2Q9_c/?t=1206) And I think it is just, yeah, working on being more ambitious, like you said. And it's funny to think about myself, you know, engineers switching from being software managers to thinking of yourself as a manager of agents. It just makes me think about the entire reason I worked as a software engineer was to avoid ever being a manager. But it's different because it's not humans. Anyways, so, yeah, I remember, I mean, Devin has been so early and you saw so many things. You made so many, you know, trends that ended up coming true. What are some more like future trends with agents? I know you've already talked about it a little bit, but where do you see things going in the future?

[20:41](https://realalpha.app/v/fFS4bs2Q9_c/?t=1241) Our focus right now really is going from individual tasks and work that an IC engineer can do to working on goals that an entire organization might work towards today, right? Usually large organizations, they're building entire products, maintaining entire applications, or sometimes just working on unsolved problems, you know, like going and like building out new technologies that don't exist yet. I don't think you can give an AI that entire problem today. But Dev and Bots, you invest in proactivity, you invest in being able to manage out to a team um and frankly you know uh you invest in things like you know better cost efficiency as well because at some scale you need to be able to take a large amount of money and efficiently allocate it to the right experiments that's one big area investment we've been working with every single model lab and you know working with the right models for the right tasks given the price performance i think that's another reason that devin bots makes a lot of sense for us right now is because people are getting to the point where they don't necessarily want to manage the model that's being used for every individual task.

[21:48](https://realalpha.app/v/fFS4bs2Q9_c/?t=1308) The bot can handle that for you, decide here are the tasks that need this model, or here's the tasks that need those models. And so I hope in a year you can give problems as large as enable new drug discovery and a dev and bot or a super dev and bot, whatever we might call it at that point, will be able to break that down

[22:10](https://realalpha.app/v/fFS4bs2Q9_c/?t=1330) **Walden Yan:** into here's experiment one, two, three, here's the different pieces of software that we have to set up to make that successful.

[22:18](https://realalpha.app/v/fFS4bs2Q9_c/?t=1338) Here's the amount that we're going to invest in these different areas. Here's different plans. And the humans can make high-level decisions like, hey, path A or path B, which ones do we want to take here? Like, oh, how much are we willing to invest and take on risk in these various experiments? But yes, it's very different from the work of what an IC engineer might look like today. I do want to quickly touch on, because I've had thoughts on this before about the changing role of an IC. I love to code. You know, I started this company in many ways because I love to code. I want to build a tool that makes coding easier.

[22:55](https://realalpha.app/v/fFS4bs2Q9_c/?t=1375) I think there was a period where we were in this uncanny valley of coding was getting a lot easier to write, or code was becoming a lot easier to write, but then you're doing the things you enjoy least, like reviewing code. um i feel us exiting that valley now i think actually devin bots is going to be a big part of this where i don't feel like i have to juggle so many threads and be reviewing the code that a different agent wrote for me i feel like i'm talking and working at this higher level now where i can actually just like let my imagination run free and i can still make the decisions i want the right architectures the right product decisions um and i get the satisfaction of clicking merge and seeing the PR merge and the confetti go off, I feel like I'm getting to a point where, wow, I'm reaching a new steady state flow that didn't exist for a little bit here in the age of AI. And that makes me hopeful that, you know, we're actually on this new ramp out where you can find this new thing to really love and enjoy about the way you code and build products today. Yeah, I completely agree.

[24:00](https://realalpha.app/v/fFS4bs2Q9_c/?t=1440) And it did feel when I was working as a software engineer that as I started using AI to code, it felt like a lot more of my job was just reviewing PRs, which again, not the most fun part. It's like coding and building something new and thinking about, thinking on these hard problems of how can I structure these things. And then it started to feel like I'm just, you know, it looks good to me on leaving that comment, like looking, sifting through someone else's code, the AI's code, which not fun at all. And yeah, I do feel like with the bots, It feels like this shift of now managing things, being super productive, less just, you know, reviewing everything in that kind of older way.

[24:38](https://realalpha.app/v/fFS4bs2Q9_c/?t=1478) Yeah, you know, I've really I've actually been pushing my bots to find more ways to lean off of my code review because that is the part that I enjoy the least.

[24:49](https://realalpha.app/v/fFS4bs2Q9_c/?t=1489) **Walden Yan:** And if you step back a little bit, the problem of how do I get my AI to not need me to review my code is actually a pretty fun engineering problem.

[24:58](https://realalpha.app/v/fFS4bs2Q9_c/?t=1498) You're thinking now less in terms of individual lines of code and debugging, but you're actually thinking about high-level guardrails. Like, oh, how can I architect my code base so that all the code over here, it's impossible to introduce security vulnerabilities? Okay, whoa, if I do that now, then all this can be auto-shipped and merged and auto-fixed by the AI, and I don't have to be worried about it. Can I add interesting new technologies like formal verification or better testing infrastructure so that I trust my agents more? If you spend more time on these meta problems, I've found that you can actually let your dev and bots run even more autonomously.

[25:34](https://realalpha.app/v/fFS4bs2Q9_c/?t=1534) And really, that's where you need to get to, to the point where you can be merging thousands of PRs without any human intervention. So actually, your early question was, where's the technology going from here? I really think that code review is going to be a thing of the past very soon. But it's not because we're just going to turn our heads away and say, I don't care to review code anymore. I think it's because we're actually going to spend time thinking about how do we engineer our code bases and set up our Devon bots so that I don't have to be in the weeds and reading every line of code. Okay, so that's a problem you're kind of currently working on is how can I review less of the code, trust my bots more to do this for me? Yeah, 100%. For example, we have Devon bots that are going in and changing the prompts for Devon continuously, but they're not just YOLO shipping them. We actually tell them any change you make, you have to run it, you know at least 10 times with like various models and you know with certain guardrails um changes beyond a certain length need human review but changes below a certain length that don't need it uh and you give it all these tools and systems where you can point in the right

[26:41](https://realalpha.app/v/fFS4bs2Q9_c/?t=1601) direction and give it the guardrails where you know it can be safe and then you get i can just turn away i like come back later and i find out oh devin like made like five or six prompt changes based off of you know human bug reports that's like an amazing feeling and that's like super rewarding that's kind of like the feeling that i i set out to do when when i came in and started coding and doing engineering yeah it's starting to feel like more fun again like i feel like i had that period as a software engineer where it's just started to feel a little depressing and now it's feeling more fun especially just the more you build and the more ambitious you try to be i think yeah the there's kind of a interesting thing of like the people who are now most ambitious i think will have a uh leg up with the tools of today and so i actually think jenny who i mentioned was running her own game studio with 20 dev and bots doing different roles and acting like you know actual uh roles of engineers at our company i was like wow i never thought of doing that or almost i was scared to push it to that level of ambition but she not only pushed it to that level of ambition she invested into making a real reality she made it happen with a

[27:50](https://realalpha.app/v/fFS4bs2Q9_c/?t=1670) real application making real revenue and users i was like wow that is that is uh fascinating and i hope more people will use this as an opportunity to, to try to do crazier things. Uh, I, I really do think that, um, it, it, it must be like a super, we, we must realize how super unique this time period is where we have these like new tools and toys that can do such powerful things, but you just have to point them in the right direction and, and, and give them the right ideas to start off what they're doing. Yeah, absolutely. And I remember some advice I was given when I was kind of having an existential crisis as a software engineer was like, just go build 10 crazy ideas. And then I think I quickly started going from scared of AI and scared of what was going to happen to getting really, really excited because it's kind of like, you know, the joke that software engineers have like, I have like 50 domains that I've never like had the time to build those projects, but they're all, I'm just hoarding those domains. And it's like, now is the time to use all those unused domains that we've all just been had in our, had for years. Yeah. I mean, this was, it feels like devon bots is realizing my original dream for devon which was i was just in my dorm room and i

[29:02](https://realalpha.app/v/fFS4bs2Q9_c/?t=1742) had like 10 20 different side projects and i wanted a ai that could just message and would go maintain those for me we ai was not nearly as good back then but we've come such a long way and i think we've built out the underlying cloud infrastructure and autonomous capabilities and multi-agent capabilities to actually make something like this possible so that's like super super satisfying um and yeah i think there's something unique too about having a devon bot that can actually go write code a lot of uh other tools out there are like personal assistants and sure they can connect to my email and run errands for me and do cool stuff i actually use a few of them um but one that i know has the capability of like a full software engineering force and i can just like throw arbitrarily large ideas at it and it won't you know fall behind And that's like, that's something that I'm really excited about with our DevonBots launch today.

[29:53](https://realalpha.app/v/fFS4bs2Q9_c/?t=1793) Yeah. And I feel like that is needed for absolutely everyone because it's kind of like you said earlier with the future of customized software and like just building any sort of app that works for you, like doesn't even have to be someone who is technical, right? I think that's great that it's good at that coding for pretty much anyone. Yeah, no, absolutely. Right on. Yeah. Okay, cool. Well, I think that's our time. Thank you so much for joining us. for someone who's just watching and they want to get started with dev and bots like what do they do first absolutely um you can just go to dev.ai slash bots uh sign up for for an account and you know play around with it awesome cool well walton thank you so much for taking the time to talk to me i've been following cognition for forever and i was really excited to talk to you so thanks so much for coming on mts thank you a huge thanks to mts sponsors blitzy autonomous software development for enterprise code bases. Ship 5X faster. Blitzy.com. AdQuick. Make your brand a billboard.

[30:53](https://realalpha.app/v/fFS4bs2Q9_c/?t=1853) Out-of-home advertising as easy to scale as digital. AdQuick.com. Arena. Measuring AI performance in the real world. Arena.ai. Support for the show comes from VCX, the public ticker for private tech. The U.S. stock market started history's greatest wave of wealth creation. From factory workers in Detroit to farmers in Omaha, anyone could own a piece of the great american companies but today our most innovative companies are staying private longer which means everyday americans are missing out until now introducing vcx a public ticker for private tech visit getvcx.com for more info that's getvcx.com carefully consider the investment material before investing including objectives risk charges and expenses this and other information can be found in the innovations fund prospectus at getvcx.com this is a paid sponsorship You
