Screenpipe: Record Your Workflow and Turn It into AI Agents
Create a personal memory for your AI assistants with screen and audio recording.
1What is Screenpipe?
Screenpipe is an application that records your screen and audio locally, creating a searchable memory for AI assistants. This makes it easier to automate repetitive tasks and create standard operating procedures. The app focuses on helping AI understand what you're doing on your computer.
2How It Works
Instead of continuous video recording, Screenpipe listens for events like app switches, clicks, typing pauses. When something meaningful changes, it pairs a screenshot with the OS's accessibility tree. OCR is used when structured data isn't available. Audio is continuously recorded and transcribed locally.
3Use Cases
Screenpipe can be used to add context to tasks, retrieve information, create a personal wiki, and discover automation opportunities. For example, you can query 'list tasks I worked on between 8am-4pm' or automatically update your CRM when visiting LinkedIn profiles.
4Privacy and Security
Screenpipe stores data locally and uses an AI PII model to redact sensitive information. It runs local models on Apple MLX or Windows DirectML. Enterprise plans let companies decide where data lives. Recording schedules and other privacy features are supported.
5Technical Details
Most of Screenpipe is written in Rust, MLX, and Onnx. It uses cider/direct C calls for Apple APIs and windows-rs for Windows APIs. Experimental Linux support exists. Desktop app and CLI tools are available. Code is source-available but commercial use requires a license.
6Why It Matters
Screenpipe enables AI to understand and automate daily work routines, boosting productivity by eliminating repetitive tasks. It's particularly valuable for teams and companies to discover automation opportunities. The tool offers a novel approach to contextual AI understanding.