Scott Wu

CEO of Cognition, which launched Devin — the first autonomous AI software engineer — and has been laser-focused on agentic coding since inception Their core argument: In AI, think stickiness rather than moats. The real defensibility is not a barrier to entry but accumulated context and learning that makes switching costly for users. The evidence is specific: Devin learning a codebase over time like a human engineer accumulating institutional knowledge. Furthermore, product surface area designed around stickiness: Slack integration, issue-based workflows, team-level adoption. Usage-based ACU pricing model aligned with the accumulated-value thesis. In Scott Wu's own words: "I'd give one slight tweak on that, which is I think it's often less about moats and more about stickiness. Moats are in some sense, typically what folks mean by moats is something that means that a competitor couldn't even enter the market. I don't think there's any kind of hard barrier that would prevent others from entering. I think what does exist is stickiness." (Reframing the AI moat debate from barriers to entry toward accumulated product stickiness.)

Built the world's first autonomous AI software engineer, used by companies from 2-person startups to Fortune 100 enterprises. His 15-person team runs with each engineer managing up to 5 Devins simultaneously, making them one of the most agent-native engineering teams in existence. Their core argument: Fully autonomous agents that work asynchronously end-to-end are the future -- the interaction model is delegation and review, not real-time collaboration. The evidence is specific: Cognition's 15-person team: each engineer uses 5 Devins, ~25% of PRs are agent-authored, expected to exceed 50%. Furthermore, devin integrates with Slack, Linear, and GitHub -- the same tools human engineers use. Failed onboarding pattern: customers who gave Devin big re-architecture tasks on day one without setup. In Scott Wu's own words: "Devin is a fully autonomous software engineer that is going to work on tasks end to end, and so there are a lot of great tools for all parts of the stack of the AI code workflow. What Devin does is it is a full asynchronous workflow, and so you can tag Devin on an issue in Slack, you're talking about an issue and you tag Devin, you can tag Devin in Linear, you can have Devin and Devin will make pull requests in your GitHub, and so it's very much built to work with engineering teams as your junior engineer." (Explaining Devin's core interaction model.)

Scott Wu is CEO of Cognition, the company behind Devin -- the AI software engineer that works with companies from two-person startups to Fortune 100 firms -- making his insistence that humans should still learn to code a counterintuitive position from someone whose product automates engineering work. Their core argument: Absolutely yes -- coding teaches structured thinking and mental models of computing that become more important as AI handles implementation. The evidence is specific: Python as an existing example of 'explaining in English what you want and the computer does it' -- from the perspective of a 1970s programmer. Furthermore, the 90/10 split of engineering time: Kubernetes errors, bug reports, and migrations versus creative problem definition and architecture. Devin being used by companies ranging from two-person startups to public banks, all of which still require human direction. In Scott Wu's own words: "My answer would be absolutely yes. I think, to a large extent, when you take computer science classes and when you learn these fundamentals, sure you're learning a little bit about how a particular language is, syntax works or something like that. But honestly, most of what you're learning really is about the ability to logically break down problems for number one. And two, I would say is just the model of a computer and a lot of these decisions and a lot of the abstractions that we've built over time." (Responding to whether people should still learn to code.)

CEO of Cognition, the company behind Devin, the first autonomous AI software engineer -- which set to write 50% of its own company's code and has been laser-focused on agentic coding since inception. Their core argument: Usage-based pricing with Agent Compute Units (ACUs) is the right model for autonomous coding agents -- pay for what the agent actually does. The evidence is specific: Devin uses ACUs (Agent Compute Units) as its billing metric for autonomous coding tasks. Furthermore, tasks are delegated via Slack, issue trackers, or direct prompts, with cost proportional to agent compute time. Devin is reportedly writing 50% of Cognition's own code, validating the autonomous agent model. In Scott Wu's own words: "We have since the beginning been laser-focused on agentic coding, and that is the one thing that we've really believed in. It's the one thing that we've designed for and that goes all the way to even the revenue model with ACUs and having the usage-based setup." (Explaining why ACU-based pricing is integral to Devin's agentic product design.)