Operating Modes
Three modes. Same architecture. Different cost.
fai's vault structure and knowledge flow never change. What changes is when synthesis runs and which model runs it.
Configure your mode with:
fai synthesize
Free. Zero AI.
Everything captured to disk. No synthesis, no AI calls. Raw vault files are searchable and inspectable. No API key needed. Free.
Managed cost. Scheduled synthesis.
Capture + AI synthesis on your schedule. Works with local models (Ollama, LM Studio, LocalAI) or cloud (OpenRouter, Groq, OpenAI). ~$1–2/day with cloud at moderate cadence.
Power mode. Always fresh.
Maximum freshness. Synthesis runs after every configured number of seals. Higher cost - immediate synthesis after each work block.
A developer using capture-only for 6 months, then switching to scheduled synthesis, immediately gets 6 months of compound intelligence synthesized. Capture is never wasted.
Available synthesis providers
| Provider | Type | Notes |
|---|---|---|
ollama | Local | Runs on your machine. Free, private. |
lmstudio | Local | LM Studio's local server. Free, private. |
localai | Local | LocalAI compatible server. Free, private. |
openrouter | Cloud | Access to many models via one API key. |
groq | Cloud | Fast inference. Low cost. |
openai | Cloud | GPT models. Higher cost, high quality. |
claude-code | Cloud | Anthropic's Claude. High quality synthesis. |
capture-only | None | No synthesis provider. Capture only. |
Setting up
fai synthesize
Runs an interactive wizard. Choose your provider, configure your API key (if cloud), and set your synthesis cadence. After that, synthesis runs automatically when enough sessions have sealed.
To re-configure at any time, run fai synthesize again.