Archive
Day 40: Why autonomy needs budgets
Day 39: Why production agents need recovery design
Day 38: Why guardrails should enable safe action
Day 37: Where human review actually belongs in AI workflows
Day 36: Why reflection should change the next action
Day 35: Why agent memory is not one database
Day 34: Why tool access is where AI becomes operational risk
Day 33: Why planning reduces wasted AI actions
Day 32: Why agent behavior is a loop, not a single prompt
Day 31: Why agents need outcomes, boundaries, and stop conditions
Day 30: Why RAG must be evaluated in parts
Day 29: Why reflection only matters when tied to evidence
Day 28: Why weak evidence should trigger recovery, not confidence
Day 27: Why freshness is part of correctness
Day 26: Why evidence is becoming multimodal
Day 25: Why production knowledge often lives in systems, not PDFs
Day 24: Why some answers live in relationships, not documents
Day 23: Why complex AI questions need decomposition
Day 22: Why one user question may need many searches
Day 21: Why RAG needs both broad context and precise evidence
Day 20: Why every important AI claim needs provenance
Day 19: Why reranking is where retrieval becomes useful
Day 18: Why relevant data can still be unauthorized data
Day 17: Why hybrid search is often the practical default