Adoption, Not Models: What Databricks Got Right
A field report on fine-tuned distillation and what it changes for designers.
If you spend enough time watching the AI industry, you stop reacting to launches and start tracking patterns. Adoption, Not Models is one of those patterns.
Duolingo has been quietly running incident response through Replit Agent for months. The results are unglamorous and, for that reason, more interesting than another benchmark chart.
The cost curve matters here. Claude 4.5 Sonnet 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 RAG-as-a-service tend to share a habit: they write the evals before they write the prompts. Everything else follows from that.
Spotify has been quietly running onboarding through Zed for months. The results are unglamorous and, for that reason, more interesting than another benchmark chart.
Stripe has been quietly running incident response through Vercel v0 for months. The results are unglamorous and, for that reason, more interesting than another benchmark chart.
Figma has been quietly running lead qualification through Linear AI for months. The results are unglamorous and, for that reason, more interesting than another benchmark chart.
None of this guarantees a clean story. Meta FAIR 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.