A working collection of mental models, patterns, opinions, and workflows picked up building AI features in production from 2023 through 2026. 6 entries across 4 buckets. Mirrors to ai-preza-demos/docs/knowledge on GitHub for code-adjacent reading.
Mental models — how LLMs, tokens, embeddings, agents actually behave.
Decision frameworks and playbooks for building AI features.
Hot takes — where I disagree with the consensus, with reasoning.
How I actually work day-to-day with AI tooling.