Back to directory

Sentient

A macOS assistant that runs a local language model every night to turn the user's files, messages and screenshots into a private knowledge base, then proactively drafts and finishes tasks the next morning.

Sentient

Sentient is an on-device AI assistant for Apple silicon Macs. Each night it reads what changed since the last run, including files, screenshots, WhatsApp, iMessage and Apple Notes read directly from the databases already on disk, plus Gmail and Google Calendar through the user's own connectors, and turns it into a plain markdown knowledge base stored on the machine.

A local model screens every item first and keeps, discards or drops it as sensitive before anything is summarized, so sensitive material leaves no summary, log or trace; only the clean, stripped summaries reach the model that builds the knowledge base. The heaviest analysis can call a frontier model through the user's own existing chatbot subscription, but Sentient holds no API key or account of its own.

By morning it prepares drafted replies and finished chores based on that context, and its Sidekick interface can be told to finish a task and then carries it out across the user's own apps. All personal data stays on the Mac, and the project is open source.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.