
Getting AI Coding Agents to solve hard problems in complex codebases
At HumanLayer, we’re on a mission to change how teams build software with AI. If we had a motto it would be “no vibes allowed” - we believe that AI agent-enabled coding is a deeply technical engineering craft, and we’re building CodeLayer to unlock it at scale. A Stanford study from June 2025 said AI can’t solve hard problems or work well in complex or legacy codebases, and we’re proving them wrong with advanced techniques inspired by Context Engineering (a term originally coined by our CEO Dex in April 2025). Learn more about our mission and learnings at https://hlyr.dev/ace-fca
Our product is an Agentic IDE built on claude code that helps CTOs and VPEs drive the transition to 99% AI-written code in their engineering orgs. Our MRR grew 100% last month, and we’re bringing on a founding engineer who loves to ship like crazy and get in the weeds with customers to solve big problems in AI Developer Productivity.
You…
We:
CodeLayer is the best way to get AI to solve hard problems in complex codebases. We help CTOs and VPEs drive the transition to 99% AI-written code in their engineering orgs.
We grew 100% in the last month, we welcomed two YC customers with 50-100 person engineering teams, and we're in paid pilots with several publicly-traded teams with thousands of engineers.
Learn more about our mission, why Stanford says AI coding agents can't solve hard problems in complex codebases, and how we're changing that in our Getting AI to Work in Complex Codebases (with over 400 comments on HN) and our talk at YC from august: Advanced Context Engineering for Coding Agents