July 23, 2026
Your laptop has receipts
Launch HN: Screenpipe (YC S26) – Power your agents by your 24/7 screen recording
This app remembers everything on your screen—and the comments are freaking out
TLDR: Screenpipe wants to give AI a memory of everything on your computer by recording screen and audio locally. Hacker News immediately turned it into a trust war, with fans calling it powerful and critics calling it creepy, risky, and one accidental overshare away from disaster.
Screenpipe showed up on Hacker News with a very big promise: an app that quietly records your screen and audio on your own computer so an artificial intelligence assistant can remember what you were doing, what you said, and even help turn your habits into repeatable workflows. Founder Louis pitches it as the ultimate “second brain,” a way to stop losing context and let software finally keep up with your real work life.
But the real action was in the comments, where readers instantly split into two camps: “this is the future” versus “absolutely not, privacy nightmare.” One of the loudest worries was painfully relatable: what happens when your work laptop also contains your personal doomscrolling, random Reddit checks, and private life? For skeptics, a tool that sees everything sounds less like productivity magic and more like inviting a nosy robot roommate to watch over your shoulder forever.
Then came the side drama. One commenter posted a warning that the site automatically downloads a .dmg file, which is the sort of sentence guaranteed to make the internet sit bolt upright. Another escalated things further by linking an accusation that the company had harvested user data without permission—instantly turning a product launch into a trust battle.
Not everyone was clutching pearls, though. A supporter said they’d already been using Screenpipe to build "AI buddies" and praised the speed. So yes: some people see a superpowered memory machine. Others see a surveillance device with a startup logo. Hacker News, naturally, chose chaos.
Key Points
- •Screenpipe is introduced as a local-only app that records screen and audio to provide searchable memory for AI agents.
- •The founder says the product emerged from years of work on personal knowledge systems, RAG experiments, Ava, and Embedbase.
- •The article argues that prior approaches such as fine-tuning, tool calling, MCP, and skills did not give AI enough day-to-day context across apps.
- •Screenpipe moved from continuous video plus frame-by-frame OCR to an event-driven approach using screenshots, accessibility-tree data, and fallback OCR.
- •Captured data is stored locally and exposed through an authenticated API, allowing built-in or external AI agents to retrieve context and summarize work.