What Your Coding Agent Can't See in the Code - Tom Moor, Linear

Categories: AI, Tools

Summary

As coding agents scale from hours to days of autonomous operation, the bottleneck shifts from code generation (now 5 cents for what cost $100 three years ago) to maintaining product coherence—requiring new systems to manage context in the outer loop rather than manual human delegation in the inner loop.

Key Takeaways

  1. AI cost-performance follows exponential decline: intelligence costing $100 three years ago now costs $0.05, enabling agents to run autonomously for hours or days rather than minutes with human oversight.
  2. The failure mode for AI-powered development shifts from inability to build to inability to maintain coherence—code production is no longer the constraint, but keeping a product coherent as a whole is.
  3. Context management moves from inner loop (tight human control) to outer loop (automated orchestration) as agents scale, requiring new infrastructure to manage multiple long-running agents in the cloud automatically.
  4. Coding agents themselves are paradoxically both mature and extremely early (emerged functionally in 2024), suggesting practitioners need to prepare for rapidly evolving capabilities while building systems flexible enough for multi-day autonomous task execution.
  5. Product quality now depends on systems that manage agent context and coordinate multiple autonomous systems, not on individual engineer productivity—shifting product development from a craftsperson model to a systems orchestration model.

Related topics

Transcript Excerpt

[music] >> Thanks for turning up to a talk about agents and context. I, you know, I I didn't think these were keywords that would really hit today. Uh, I'm sure it won't be the the last the last time you hear about it. But I'm going to try and kind of get out of the the inner loop a little bit uh and think more about the outer loop of agents. So, um excuse me if it feels a little off-topic at times, but we'll get there. So, yeah, I I work at a company called Linear. Um I'll start with a just a quick intro for those in the room that might not be familiar. Uh Linear started life as an issue tracker. Um it was born out of the frustrations with the existing products in the space. Uh how poor quality they were, kind of slow, hard to use, buggy, slow. I won't name any specific products, but you …

More from ai.engineer