Inflection's Latest Move Signals a Shift Toward evals-first development
Notes from the teams shipping voice mode to real users.
The interesting question is not whether Notion AI works. It does. The interesting question is what teams do with it once the novelty wears off.
Inside Duolingo, 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 RAG-as-a-service 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 long context is exactly the kind of feature that becomes a platform when nobody is paying attention.
Eval harnesses, once an afterthought, are becoming the most important piece of code in many AI projects. Stripe's team treats theirs the way an SRE team treats a runbook.
The cost curve matters here. Grok 4 is roughly an order of magnitude cheaper per token than the equivalent model 18 months ago, and that changes which problems are worth automating at all.
Teams that win with small-model orchestration tend to share a habit: they write the evals before they write the prompts. Everything else follows from that.
None of this guarantees a clean story. xAI 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.