Cognition’s Scott Wu: AI coding agents augment, don’t replace developers

At a glance:

• Cognition raised $1 billion at a $26 billion valuation, led by CEO Scott Wu.
• Devin, the AI coding agent, handles 89% of code commits at Cognition, with Windsurf agents managing the remainder.
• Wu emphasizes AI agents should enhance human creativity, not replace developers, citing personal programming experience and industry trends.

What happened

Cognition, the two-year-old AI coding agent startup behind Devin, announced a $1 billion funding round at a $26 billion valuation, marking a significant milestone in the AI-driven software development space. In a blog post accompanying the raise, the company outlined its vision of "self-driving software development," positioning Devin as a tool to automate routine coding tasks. CEO Scott Wu clarified that while Devin can operate independently, it functions at a level comparable to a junior-to-mid-level engineer depending on the complexity of the task. The company also revealed that 89% of code committed by its engineers was generated by Devin, with the remaining 11% handled by local agents from Windsurf, a competitor it acquired in 2025.

The vision for self-driving software

Wu envisions AI agents like Devin as a new layer of abstraction in software development, similar to how visual programming environments shifted focus from machine code to higher-level logic. He argues that these tools can handle the "long-tail maintenance tasks" that developers often find tedious, such as updating legacy systems or migrating applications across platforms. This approach aims to free human programmers from repetitive work, allowing them to focus on creative problem-solving and product innovation. However, Wu stops short of predicting full autonomy, noting that the industry is still in the early stages of understanding recursive AI systems that improve themselves over time.

Human-centric approach

Despite the automation narrative, Wu insists that Devin was never designed to replace human developers. He highlighted his own background as a competitive programmer, starting at age nine, and his belief that coding is fundamentally about creativity and turning ideas into tangible products. To underscore this philosophy, he keeps a stuffed toy on his desk—a physical representation of Devin as a collaborative "buddy" rather than a replacement. Wu emphasized that software engineers derive joy from building, and AI should preserve that experience rather than diminish it.

Industry implications

The rise of AI coding agents has sparked debates about job displacement, particularly in 2026 as tech companies increasingly cite AI for workforce reductions. Wu’s stance contrasts with this trend, advocating for augmentation over replacement. He pointed to parallels in other industries, such as customer service and medicine, where AI could similarly assist professionals without eliminating their roles. While acknowledging the potential for disruption, Wu stressed that human agency must remain central to decision-making in all fields, not just software engineering.

Future outlook

Wu predicts that AI agents will expand beyond coding into other sectors, driven by the same principles of task automation and human collaboration. He sees this as part of a broader shift toward "recursive" AI systems that learn and adapt, though he admits the trajectory is uncertain. For now, Cognition’s focus remains on refining Devin’s capabilities while ensuring it aligns with developers’ needs. The company’s rapid growth and high valuation suggest strong investor confidence in this vision, even as questions linger about the long-term impact of AI on the workforce.

Why it matters

The debate over AI’s role in coding reflects a larger tension in the tech industry: balancing efficiency gains with ethical considerations. Wu’s perspective offers a middle ground, suggesting that AI can enhance productivity without erasing human contributions. This approach could influence how other companies integrate AI tools, potentially shaping policies around job security and technological adoption. As AI becomes more pervasive, stories like Cognition’s will likely define the narrative around human-AI collaboration in the coming years.