May 2, 2024 · 29m · no-priors
No Priors Ep. 62 | With Cognition CEO and Co-Founder Scott Wu
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Cognition CEO Scott Wu discusses the creation and technical design of Devin, the autonomous AI software engineer, while exploring how competitive programming principles and agentic AI are transforming the future of software development.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 23.1% of the talking time here. How this is scored →
The hosts as informed peer 4.3 Guest teaching 4.0 Guest disagreement 0.3 The hosts pushing back 0.4
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 22:49 Pushback against English replacing CS fundamentals
Scott reframes the popular industry meme that natural language is the only necessary programming language, insisting that understanding low-level systems like TCP and logic gates remains vital.
Hardest push from the hosts ▶ 24:06 Elad questions the 20-year outlook for studying computer scienceElad pushes back on conventional career advice, questioning whether studying math and computer science will remain practical if machines handle autonomous planning.
Biggest teaching moment ▶ 17:20 Why agents require interactive debugging over direct predictionScott educates the hosts on why one-shot diff generation fails on complex codebases compared to human-like execution loops involving logging, testing, and stack trace inspection.
The host holds their own ▶ 12:08 Elad evaluates agent steering paradigms and failure modesElad showcases his deep product grasp of AI agent architectures by comparing existing autonomous agents to wandering interns and breaking down the necessity of Devin's multi-pane visibility.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Early Coding Journey and the International Olympiad in Informatics | 3 | 4 | 0 | 0 | Sarah inquires about competitive programming practice and whether problem sets are fixed algorithmic domains or open-ended challenges. Scott explains the training mindset and how IOI problems require novel reductions to core algorithms. | |
| Online Subcultures, First Principles, and Tech Entrepreneurship | 4 | 3 | 0 | 0 | Elad and Sarah draw connections between early internet communities, competitive programming circles, and Silicon Valley founders. Scott elaborates on first-principles thinking and questioning baseline assumptions. | |
| Founding Cognition and the Genesis of Devin | 3 | 5 | 0 | 0 | Elad and Sarah prompt Scott on Cognition's inception and Devin's core autonomous capabilities. Scott educates listeners on Devin's end-to-end command-line and browser execution model. | |
| Viral Public Reactions and the Future of Engineering Work | 5 | 4 | 1 | 1 | Sarah asks whether software engineering will disappear as a career path, offering the historical analogy of human calculators to argue for latent demand. Scott agrees, explaining that shifting time from syntax writing to system architecture expands developer output. | |
| Devin's Workflow, UI Paradigm, and Core Strengths | 6 | 5 | 0 | 0 | Elad demonstrates domain expertise by analyzing UI failure modes of early AI agents, contrasting unguided execution with Devin's steerable multi-pane interface. Scott explains where Devin excels, such as complex DevOps tasks, versus where human guidance remains essential. | |
| Technical Foundations, Agent Reasoning, and Industry Trajectory | 6 | 5 | 0 | 0 | Sarah and Elad probe the boundary between foundation models and agent scaffolding, asking about memory, self-play, and reasoning. Scott explains why iterative execution, debugging, and tool usage outperform one-shot code generation. | |
| Long-Term Evolution of Software Engineering and Core Fundamentals | 5 | 4 | 1 | 2 | Elad questions whether advising future generations to study computer science still makes sense over a 20-year horizon. Scott pushes back against the notion that fundamentals are obsolete, emphasizing the persistent value of systems knowledge and low-level logic. | |
| Cognition Team Culture and Talent Acquisition Philosophy | 2 | 2 | 0 | 0 | Sarah and Elad ask about Cognition's hiring strategy and talent profile. Scott outlines their emphasis on high-ownership generalists and founder-caliber engineers. |
Statements from this episode (10)
Scott Wu: Many IOI competitive programming peers became founders in AI
“No, it was a very, very tight knit group. A lot of, honestly, a lot of that group, all, many, many of them became founders. And a lot of them went into AI.”
Scott Wu: Great startups, like math competitions, require identifying non-intuitive truths
“A lot of the really great math or programming problems are kind of about, you know, what are all these assumptions that you take for granted and which one of those might actually not be true, you know and a lot of these really tough problems and really interes…”
Wu: Devin autonomously makes decisions and executes engineering tasks end-to-end
“Devin is, is fully able to make all of its own decisions in the same way that a human software engineer would. And so you know, obviously writing and editing code is a huge piece of that, but also, you know, being able to work with the command line, being able…”
Scott Wu: AI will lead to more software engineers, not fewer
“I think if anything, there's going to be more engineers, not less and I'll kind of give two reasons why. I think the first is there's just so much demand for engineering out there and so much demand, honestly, even in ways that we don't always think about, rig…”
Scott Wu: AI will enable software engineers to do 5x more
“I think the average software shares today might spend 20% of their time thinking through all these, like, really fundamental problem solving questions, and 80% of the time writing that in code. And I think they'll, they'll be able to do five x more, and they'l…”
Wu: Cognition builds Devin using Devin itself internally
“We ourselves use Devon all the time in our own work, even when we're building Devon, working through each of the pieces, basically, and figuring out what it took from us to you know, to work well with Devon.”
Wu: Devin executes precise specifications rather than deciding product direction
“Devin is very much not the one who is going to be deciding what to do. If that makes sense. And so, you know, we very much think of Devin as you provide the precise formulation of what you want built or what you want done, and Devin is the one that is doing th…”
Wu: In 10 years, learning coding languages will seem obsolete
“You know, I think 10 years from now, people will look back and think, wow, isn't it crazy that, you know, you had to learn all these esoteric languages and you had to, you know, work through all these stack traces or whatever just to be able to communicate wit…”
Wu: Future software engineers will resemble technical architect and product manager hybrids
“I think the, you know, I think the role of A software engineer five or 10 years from now, it looks something like a mix between a technical architect and a product manager today, you know, where, where a lot of what you do is, you know, you take problems that …”