Archive
Day 15: Why vector stores are infrastructure, not magic memory
Day 14: Why embeddings are useful and easy to overtrust
Day 13: Why chunk boundaries shape answer quality
Day 12: Why retrieval quality starts before search
Day 11: Why RAG is an evidence design problem, not a buzzword
Day 10: When to use a chatbot, workflow, or agent
Day 09: Why a capable model is still not a product
Day 08: Why production AI needs structured outputs
Day 07: Why confident AI answers still need evidence
Day 06: Why creativity settings are product decisions
Day 05: Why vague prompts become vague systems
Day 04: Why more context can make an AI system worse
Day 03: Why tokenization quietly affects cost, limits, and reliability
Day 02: Why LLMs feel intelligent but still generate one step at a time
Day 01: Why model choice is rarely the first production AI problem