Vellum is a personal AI assistant that you configure with your own context and personality and then reach from any of your devices. It keeps a persistent memory of people, preferences, and ongoing work across conversations, and it takes actions on your behalf rather than only answering questions.
The assistant can browse the web, manage files, run code, handle calendars and messages, and call third-party apps. Routines and heartbeats let it work while you are away and notify you of what matters, so tasks such as chasing invoices, enriching leads, or drafting a deck can run in the background.
The same assistant is available on the web, macOS, iOS, the command line, email, Telegram, and Slack. The runtime is open source and can run on Vellum Cloud or on your own machine, and an assistant can hold its own identity with a separate email address, GitHub account, and Slack handle.
Features
- Persistent memory: context about people, preferences, and projects carries across conversations over months
- Configurable persona: you set the assistant's personality and supply the context that makes it yours
- Actions: web browsing, file management, code execution, calendar, messaging, and app integrations
- Routines and heartbeats: scheduled and background work with notifications on completion
- Self-improving skills: workflows the assistant learns and reuses over time
- Multi-channel access: web, macOS, iOS, terminal, email, Telegram, and Slack
- Own identity: a dedicated email address, GitHub account, and Slack handle per assistant
- Deployment choice: hosted on Vellum Cloud or run locally with the open-source runtime
- Data handling: conversations are not used for model training and data is exportable and deletable
