screenpipe
An open-source local tool that continuously turns your screen and audio into a searchable memory layer, helping heavy computer users produce retrievable work history, search results, and automation context.
Tool overview
Based on the available evidence, screenpipe is worth considering as an early but promising local memory infrastructure tool, not a polished plug-and-play assistant. The heat proof is strong: 20k+ GitHub stars and multiple high-reach X posts show clear attention. But proof of usefulness is narrower and comes mainly from the official repo plus a small number of hands-on user reports, so the sample is still limited.
Its practical role is closer to a local computer-activity memory layer than a normal screen recorder, and it is not just a generic AI chatbot. From the repo and user descriptions, it continuously captures screenshots for OCR and audio for transcription, then stores what you saw, heard, or said as searchable context. That context can feed Obsidian, Notion, or agent workflows for recall, SOP generation, and work-log reconstruction. A better analogy is an open-source, controllable alternative to Windows Recall or Rewind.
On cost and setup, the evidence only supports that the open-source project is free to use and that the official desktop app has free and paid tiers without publicly confirmed pricing.