The substrate wiki.
Apiary's docs are LLM-seeded, founder-verified, and re-read from the codebase every time. No stale docs. No tribal knowledge lost between sessions. Verified pages are canonical; candidate pages are drafts awaiting Austin's stamp.
Ask the substrate
Type a question. The LLM reads the docs + the code and answers in Apiary's voice. (Coming soon — v0 routes to chat with a docs-mode system prompt.)
Foundations
What Apiary is, why it exists, the core idea.
BYO-LLM, explained
CandidateWhy Apiary refuses to bake in an LLM provider, and the five paths a user can take.
The substrate
CandidateWhy Apiary is markdown-first and brain-region-modular all the way down.
What is Apiary?
CandidateThe one-paragraph answer plus the six-pillar overview.
Six Pillars
Cortex, Committee, Cater, Reconsider, Oversee, Situational — the deep read.
Cater — the symbiotic developer-agent loop
CandidateWhy "folder = lobe, file = memory, README = identity" is the entire programming model.
Committee — House-of-Reps governance per sector
CandidateWhy a legislature beats a hierarchy for agent organizations.
Cortex — brain-region modular architecture
CandidateThe biomimetic case for treating each agent capability as a specialized brain region.
Oversee — the founder as constitutional overseer
CandidateThe right relationship between the human and a self-governing agent organization.
Reconsider — self-correction via revote
CandidateHow Apiary unsticks itself from its own past decisions without thrashing.
Situational — felt-sense judgment over rule-following
CandidateWhy retry counts and timeouts are not the right done-detection signal.
LLM Knowledge
Cloud vs local, fine-tuning, OAuth, the underlying machinery.
Cloud vs local LLM — the honest tradeoff
CandidateAnthropic vs Ollama isn't religion — it's tradeoffs across five axes.
Fine-tuning, explained
CandidateWhat training data, GPU rental, and "fine-tune the model" actually mean in dollars and hours.
OpenRouter + OAuth — one key to rule the providers
CandidateWhy OpenRouter is the lowest-friction path to "try every model with one credential.
Apiary Operations
Founder console, session budgets, foundation team — how the hive runs.
The Foundation Team — Apiary's data pipeline crew
CandidateThe 6-agent team that maps assets, builds the foundation DB, and bootstraps every later phase.
The founder console
CandidateThe 7-tab admin surface and the design philosophy behind each tab.
Session budgets — the fuel-up-and-go pattern
CandidateWhy bounded sessions beat open-ended agent runs for indie cost control.