learn
Local-first AI — sovereign retrieval that never phones home
dig runs fully on your machine. With no configured endpoint it makes zero network calls, your data never leaves the device, it is self-hostable, and there is no vendor lock-in.
Retrieval this good usually implies a hosted vector service. dig inverts that: the store, index, policy engine, and dedupe all run on-device, so your knowledge base stays yours.
Zero network by default
dig's deterministic core does the whole job with no model and no cloud. AI is an optional layer, off by default, and even when on it points at an OpenAI-compatible endpoint that defaults to localhost.
- With
mode = offor a localhost endpoint, dig makes zero external network calls. - No cloud account, no telemetry, no required network.
- The only optional network hop is an LLM or embedding endpoint you configure.
dig find "renewal terms" --mode fts
FTS needs no model at all, so deterministic search always works offline.
Self-hostable, no lock-in
dig is a single static pure-Go binary — no cgo, cross-compiles everywhere, runs offline. The AI layer speaks one OpenAI-compatible contract, so the endpoint is just a base_url: a local runtime like Ollama or llama.cpp, or a gateway you control. Swap it freely; nothing ties you to a vendor. The benchmark scores — 98.0% hit@5 on LongMemEval — were produced exactly this way, on CPU with small open models.