Cognition Chief Executive Officer Scott Wu Clarifies Artificial Intelligence Agents Serve as Accelerators Not Substitutes
Cognition CEO Scott Wu has highlighted that AI coding agents function as developer helpers rather than replacements. This emphasis underlines the crucial value of human-in-the-loop oversight.
Why it matters
You should focus your efforts on defining high-level software architecture and review boundaries, letting subagents handle repetitive boilerplate implementation.
TL;DR
- 01Structure your agent commands around highly specific sub-tasks instead of giving wide-open instructions.
- 02Establish mandatory code review gates for all files produced by autonomous coding engines.
- 03Focus on system design and boundary definition, leaving the boilerplate execution to your agents.
Human-AI Collaboration
Scott Wu, CEO of Cognition, emphasizes that their agent 'Devin' is intended as a tool for developers, not a replacement. He describes the agent as a 'buddy that helps you build more.' The company views agents as a new layer of abstraction between envisioning software and producing it.
Current Agent Performance
Wu characterizes Devin’s performance as falling somewhere between a junior and a mid-level engineer depending on the task. Currently, the company reports that 89% of code committed internally is written by Devin.
Future Vision
Cognition aims to free developers from 'toil'—the tedious maintenance and platform migrations that programmers generally dislike. Wu stresses that the ultimate decision-making authority must remain with humans across all industries where AI agents are applied.