Screenpipe records what appears on your screen and what you hear, turns it into local searchable history, and lets approved AI agents use that history as context. It captures accessibility data, screen frames with OCR fallback and audio transcripts, stores them in a local database, and exposes a local REST API and an MCP server. Scheduled workflows called pipes can write meeting notes, update tickets or draft follow-ups. Raw history stays on the device by default. A free plan covers one device, Basic costs $21 per month and Business $42 per seat per month, with Enterprise on custom terms.
It suits founders, sales and customer-facing staff, consultants and engineers who want an assistant that remembers what actually happened in their day. IT and security teams can roll it out across a company with MDM, SSO, audit logs and retention controls, while privacy-conscious users exclude password managers, windows, apps or URLs and redact sensitive fields before storage. It runs on macOS, Windows and Linux and the source code is available to inspect. Students, researchers and faculty get 50% off the first month of an individual plan. Budget for storage of roughly 30 GB per month and 600 MB to 4 GB of RAM.
Typical uses include asking what you told a client about pricing last month, summarizing the last five meetings with an account, and drafting a weekly update from calls and screen activity. A pipe can trigger when a meeting ends, write a summary with action items per person and save it as Markdown in a notes folder. Others update Linear tickets, push data to HubSpot, build a contact dossier or generate standard operating procedures from recorded work. Connections cover Linear, Slack, Notion, GitHub and Obsidian, and agents can query the local API from a terminal.
