compare
dig vs hosted RAG-as-a-service — local-first ownership over a managed pipeline
RAG-as-a-service runs a managed cloud pipeline that ingests your documents and answers over them. dig is a local-first retriever you own — hybrid search, reversible history, no data leaving your machine.
Same job — retrieve over your knowledge — opposite trust models.
What each one is
Hosted RAG-as-a-service is a managed cloud pipeline: you upload documents, the service chunks, embeds, and indexes them, and you query over an API. The infrastructure is operated for you, and your data lives on the vendor's platform.
dig is a local-first binary you drive from the command line or over MCP. It indexes a directory of real files and retrieves with a hybrid FTS + vector pipeline, with every change journaled and reversible. Nothing leaves the machine.
Where they differ
- Location. A hosted service runs in the vendor's cloud on your uploaded data. dig runs fully local, on CPU — your knowledge base never leaves.
- Operation. A managed pipeline removes setup but adds a dependency, an API, and a bill. dig is a single binary;
dig mcpis the whole integration. - Reversibility. dig journals every mutation and
dig undowalks it back. A hosted index gives you no changeset history. - Portability. dig is harness-agnostic — any agent (Claude Code, Codex, Cursor) drives it over MCP, with no lock-in to one vendor's API.
When to use which
Reach for hosted RAG-as-a-service when you want a managed pipeline with no infrastructure to run and you're fine with your documents living in the vendor's cloud.
Reach for dig when ownership matters — local data, reversible history, no service, and the same hybrid retrieval available to any agent.