Vellum Assistant

A personal AI assistant that learns how you work, remembers what matters, and takes action across your apps. It is MIT-licensed, built by Vellum, and designed to work out of the box instead of being hatched repeatedly like other personal agents.

Episodic, semantic, procedural, emotional, prospective, behavioral, narrative, and shared. Each has its own staleness window, hybrid dense and sparse retrieval, and per-user and per-channel isolation, with embeddings running locally by default rather than in a file you maintain.

Download the macOS or Windows app from vellum.ai/downloads, or use the CLI: bun install -g vellum followed by vellum hatch. The CLI supports wake, sleep, client, ps, terminal, and upgrade, and all commands target the default assistant unless you pass an ID.

Managed mode signs in through Vellum Cloud and needs no local runtime, while Local mode runs everything on your machine. The README says both use the same codebase and the same data model.

Actor identity (guardian, trusted, unknown) is resolved once and enforced everywhere; unknown actors cannot read memory, trigger tools, or escalate. Credentials live in a separate process and never reach the model, every tool call runs in a sandbox, and the default is to deny.

Channels include macOS, Windows, iOS, web, voice, email, Telegram, Slack, and Twilio. Built-in OAuth covers Slack, Notion, Google, HubSpot, Linear, Discord, Twitter, Telegram, and Twilio, and models can come from Anthropic, OpenAI, Gemini, Fireworks, OpenRouter, MiniMax, or OpenAI-compatible endpoints.

The digest does not say so; the README covers memory, identity, proactivity, security, channels, skills, and hosting without mentioning Jev or TypeSafe. It reached this catalog through the logicrw community list.