Why structured outputs Beats Bigger Models for Real-World Agents
Notes from the teams shipping vision input to real users.
The interesting question is not whether Replit Agent works. It does. The interesting question is what teams do with it once the novelty wears off.
Inside Linear, the rollout looked less like a moonshot and more like a slow migration. A pilot, a champion, a quiet expansion, a budget line.
Teams that win with tool-first agents tend to share a habit: they write the evals before they write the prompts. Everything else follows from that.
Teams that win with evals-first development tend to share a habit: they write the evals before they write the prompts. Everything else follows from that.
The skeptical read is that we are watching a feature, not a platform. The optimistic read is that memory is exactly the kind of feature that becomes a platform when nobody is paying attention.
The skeptical read is that we are watching a feature, not a platform. The optimistic read is that vision input is exactly the kind of feature that becomes a platform when nobody is paying attention.
The skeptical read is that we are watching a feature, not a platform. The optimistic read is that voice mode is exactly the kind of feature that becomes a platform when nobody is paying attention.
None of this guarantees a clean story. Google DeepMind could ship a model next month that rearranges the assumptions in this piece. But the direction of travel, for now, is clear enough to plan around.