The archive · AI & Models · Product decision · 2024-2026
Screenpipe bets agents need local memory of everything you see and hear
A solo developer's local screen-and-mic recorder hit #1 GitHub trending and HN's front page, raised seed money, and joined YC S26.
Screenpipe
What the business is
An open-source, local-first desktop app that continuously records your screen and audio, indexes it (accessibility data, OCR, speech transcription), and exposes that history to AI agents like Claude, Codex, and Cursor via a local API and MCP server.
Starting capital：Self-built by the founder from 2024; ~$2.8M seed raised by July 2025 (project milestone), then YC S26 (2026).
How it started
Louis Beaumont started Screenpipe in 2024 as a personal CLI to record his own screen and feed that context into AI. He had kept a "second brain" since 2020 and built earlier AI tools (the Ava Obsidian plugin, then the Embedbase RAG API), which taught him that models need context about what you are actually doing. An HN user posted the raw project on 2024-09-30, and the 125-comment discussion shaped it: consent, local security, CPU use, and whether agents could act on the data.
What happened
The post went viral - #1 on GitHub trending, HN front page, ~5,000 stars overnight. Beaumont turned the CLI into a desktop app with event-driven capture (accessibility tree plus OCR fallback), local Whisper transcription, a REST API and MCP server, and "pipes": scheduled agents written as markdown files. He raised ~$2.8M in seed funding by July 2025, moved to a source-available commercial license (free for personal, non-commercial use), and joined YC S26, launching publicly on 2026-07-23 with pricing from $25/month and enterprise team plans.
How it ended up
Still live and scaling: the app runs on macOS, Windows, and Linux; the repo holds 17k+ GitHub stars; and Screenpipe positions itself as the leading source-available alternative to Rewind/Limitless, Microsoft Recall, and Granola, with 16+ pipes and 45+ app integrations for teams.
Background
Screenpipe is a local-first desktop app that continuously records what you see and hear on your computer and turns it into a searchable, AI-readable memory. Louis Beaumont started it in 2024 as a personal CLI after years of building "second brain" tools, convinced that AI agents keep failing not on model quality but on context: they do not know what you were doing five minutes ago.
The bet was that users would accept an always-on screen-and-mic recorder if the data stayed on their own machine, and that developers would want to plug that memory into agents they already use. Screenpipe exposes everything through a local API and MCP server, so Claude, Codex, Cursor, or any agent can query "what did I see in the last five minutes" and act on it, while accessibility-tree capture, OCR fallback, and local Whisper transcription keep CPU and storage low.
The project went viral on 2024-09-30: #1 on GitHub trending, HN front page with 218 points and 125 comments, roughly 5,000 stars overnight. Beaumont raised ~$2.8M in seed funding by July 2025, moved to a source-available license (free for personal use), and joined YC S26, launching publicly on 2026-07-23. As of September 2026 Screenpipe is scaling with paid tiers from $25/month and enterprise team plans, positioning itself as the auditable alternative to Rewind/Limitless and Microsoft Recall.
What has to be true
- The missing layer was real: agents had tools and models but no record of what you actually did, so local memory made existing assistants more useful instead of competing head-on with them.
- Privacy as distribution: by keeping data 100% local and source-available, an app that records everything neutralized the surveillance objection that would have killed a cloud version.
- The HN and GitHub audience became product managers: the 125-comment 2024 thread steered consent, local security, CPU usage, and actionability - feedback that shaped the event-driven capture design.
- A solo, personal project shipped fast: earlier tools (Ava, Embedbase) narrowed the scope to the context layer, and the CLI-first approach let technical users spread it before any marketing existed.
What can be applied
A personal pain point can be a platform wedge: give existing agents the context they lack, and make trust the product - "your data never leaves your machine" is why an always-on recorder got adopted.
Aftermath
As of September 2026 Screenpipe is live and scaling. The repo has 17k+ GitHub stars, the desktop app runs on macOS, Windows, and Linux with subscription tiers starting at $25/month, and the team sells an enterprise plan with central config, SSO, and per-pipe AI data permissions. It maintains an MCP server and 16+ published pipes (meeting summaries, standups, CRM sync, time tracking) and positions itself as the leading source-available alternative to Rewind/Limitless, Microsoft Recall, Granola, and cloud-based screen-AI rivals like Littlebird.
Sources
- screenpipe/screenpipe - YC (S26) | Open Computer History
- Screenpipe: 24/7 local AI screen and mic recording
- Launch HN: Screenpipe (YC S26) - Record how you work and turn that into agents
- screenpipe vs Littlebird: local-first vs cloud screen AI
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card