Your personal AI memory
screenpipe records what you see and hear on your computer, creating a searchable memory you can query with local AI or approved cloud workflows.
Quick answer
Turn your computer into a personal AI memory. screenpipe records your screen and audio 24/7, creating a searchable, local-first second brain.
Your brain wasn't built for this
You process hundreds of pieces of information every day. Your brain forgets most of it.
You saw something important on screen last week but can't remember where
Meeting decisions and action items slip through the cracks
You waste time re-searching for information you've already found
Context-switching between apps means losing track of what you were doing
Your notes only contain what you remembered to write down, a fraction of what you consumed
A second brain that never forgets
screenpipe captures your screen and audio continuously, extracts text via accessibility APIs and transcribes audio, storing everything in a local database. Ask it anything about your digital life. It remembers what you don't.
Total recall
Every screen change captured and text extracted. Every conversation transcribed. Nothing slips through.
AI-powered search
Ask questions in natural language: 'what was that API endpoint?' or 'what did Sarah say about the budget?'
Local-first and private
Store memory locally first, keep the code auditable, and decide separately whether to enable sync, exports, connectors, team workflows, or cloud AI.
Works with any AI
Use Claude, GPT, Ollama, or Apple Intelligence to query your memory. Your choice of model.
How it works
Install and forget
Download screenpipe, grant permissions, and let it run in the background. It captures your screen and audio automatically. No manual action needed.
Build your memory
Over hours and days, screenpipe builds a comprehensive, searchable index of everything you've seen and heard. The longer it runs, the more valuable it becomes.
Search and ask
Use the search interface, the REST API, or AI chat to find anything from your history. Ask questions like 'what did I work on yesterday?' and get structured answers.
Code examples
Search your memory
Find anything you've seen on screen or heard in a conversation.
# search screen text
curl "http://localhost:3030/search?q=api+endpoint&content_type=ocr&limit=10"
# search audio transcriptions
curl "http://localhost:3030/search?q=budget+meeting&content_type=audio&limit=10"
# search by app and time range
curl "http://localhost:3030/search?q=design&app_name=Figma&start_time=2026-02-17T09:00:00Z"Ask AI about your day
Use natural language to query your memory through any AI model.
curl http://localhost:3030/ai/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-4",
"messages": [
{"role": "user", "content": "what did I work on yesterday afternoon?"}
]
}'Build a daily digest
Automatically summarize your day and send it to your notes app.
const today = new Date().toISOString().split("T")[0];
// fetch today's screen activity
const screen = await fetch(
`http://localhost:3030/search?content_type=ocr&start_time=${today}T00:00:00Z&limit=100`
).then(r => r.json());
// fetch today's conversations
const audio = await fetch(
`http://localhost:3030/search?content_type=audio&start_time=${today}T00:00:00Z&limit=50`
).then(r => r.json());
// generate summary with AI
const summary = await fetch("http://localhost:3030/ai/chat/completions", {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({
model: "gpt-4",
messages: [{
role: "user",
content: `Summarize my day based on this data:
Screen: ${JSON.stringify(screen.data.slice(0, 30))}
Audio: ${JSON.stringify(audio.data.slice(0, 20))}`
}]
})
}).then(r => r.json());Key benefits
Frequently asked questions
About 5-10 GB per day with default settings. screenpipe uses efficient compression and you can set retention limits to automatically delete old data. Most users keep 1-3 months of history.
No. screenpipe uses hardware-accelerated encoding and typically uses less than 1% CPU. It's designed for 24/7 use without impacting performance.
Yes. All data is stored in a local SQLite database on your device. Nothing is sent to any server. The code is open source (MIT license) so you can verify this yourself. Over 16,000 developers have reviewed the code on GitHub.
Yes. screenpipe supports offline capture, text extraction, and transcription. You can use local AI models such as Ollama or Apple Intelligence for search and summaries without an internet connection.
screenpipe is open source, works on Mac/Windows/Linux (not just Copilot+ PCs), captures both screen and audio (Recall is screen-only), and has a developer API. Recall is free on compatible hardware but limited in scope.
Yes. You can filter searches by app name, time range, content type (screen or audio), and more. The API supports all these filters, and the search UI makes it easy to narrow down results.