Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
An on-device LLM understands your entire life, then proactively offers to get your work done through computer use.
| Date | Stars |
|---|---|
| 2026-07-31 | 391 |
| 2026-08-04 | 409 |
| 2026-08-06 | 409 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
35.0
growth rate 0.00%/day
<div align="center"> <a href="https://sentient-os.ai"><img src=".github/readme/hero.jpg" alt="Your Mac, sentient. An on-device LLM understands your entire life, then proactively offers to get your work done through computer use. Visit sentient-os.ai." width="820" /></a> <samp>open source · free · optimized for Apple silicon</samp> </div> <br/> Every AI you use today has two problems: it knows nothing about you (nobody can paste their entire life into a chat box), and it only helps when summoned. Fixing both means reading your entire life, every day. In the cloud, that much inference is a fortune and a privacy nightmare. On your own chip, it's free, and nobody else's business. So every night, Sentient's on-device LLM reads what's new (your files, screenshots, WhatsApp, iMessage, Notes, email) and distills it into a clean markdown knowledge base: the deepest memory of you any AI has ever had. The last 10% of compute taps a frontier model through your own ChatGPT subscription; nothing ever touches a Sentient API key, because there isn't any. By morning, Sentient offers your grunt work done in one click: that reply you forgot, drafted from your own rich personal context. The subscription you meant to cancel, caught the night before it renews. And anywhere on your Mac, click the notch and tell Sidekick: "Finish this for me". It uses your apps like you would, even in the background while you carry on. No accounts, and your raw personal data never leaves your Mac. Free, forever. And every word of this is verifiable right here in this codebase :) <br/> ## While you sleep, it reads your entire life. <sub><samp>3:00 AM · ON THIS MAC · NOTHING LEAVES</samp></sub> At 3 AM every night, Sentient quietly wakes your Mac (lid closed is fine; it falls back asleep after) and reads what's new: files and screenshots, WhatsApp, iMessage, and Apple Notes, decoded straight out of the databases already sitting on your disk. Gmail and Google Calendar ride in through your own OpenAI connectors. Every single item passes an on-device bouncer. Gemma 4 E4B, running locally, reads it and rules: keep, junk, or sensitive. Junk gets dropped. Sensitive gets dropped harder: no summary, no log, no tombstone, zero trace it ever existed. Only clean, PII-stripped summaries of the keepers survive. Then the finale: your own codex's frontier model takes those thousands of PII-stripped summaries and distills them into the best possible knowledge base: an Obsidian-style folder of plain markdown, on your Mac, yours to read, edit, and delete note by note. It's not a black box; it's a folder. <div align="center"><img src=".github/readme/constellation.gif" alt="The Knowledge window's Constellation View: your life, from above. Notes as stars, wikilinks as threads." width="780" /></div> <br/> ## So you wake up to your work finished in one click. <sub><samp>9:00 AM · PROACTIVE INTELLIGENCE</samp></sub> <div align="center"><img src=".github/readme/morning.jpg" alt="The morning cards: replies drafted, plans researched, one tap from done" width="780" /></div> Overnight, a frontier model reads the night's findings against everything it knows about your life and prepares the few things really worth doing. The reply you forgot, drafted from your own rich personal context. The subscription you meant to cancel, caught the night before it renews. You get a small handful of cards drafted in the morning, which you can read, edit it if you like, and click once to fire. That click is the only thing that ever fires an action. <br/> ## One more thing. Meet Sidekick. <sub><samp>HOLD RIGHT ⌘ · ANYWHERE</samp></sub> <div align="center"><img src=".github/readme/sidekick.gif" alt="Sidekick: the notch drops open and does the task in your own apps" width="780" /></div> Anywhere on your Mac, click your notch (or use right ⌘: hold to speak, tap to type) and say: "finish this for me", "reply with the update from Sarah", "put these ingredients in my cart". The notch drops open
Excerpt of 18,209 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:4c06e941e0d7889b, llm:Topics: ai-agents, intelligence-layer, knowledge-base, llm, on-device-ai, personal-ai, proactive-intelligence, second-brain; description: 'An on-device LLM understands your entire life, then proactively offers to get your work done through computer use.'
matched fp:4c06e941e0d7889b, llm:Topics: ai-agents, intelligence-layer, knowledge-base, llm, on-device-ai, personal-ai, proactive-intelligence, second-brain; description: 'An on-device LLM understands your entire life, then proactively offers to get your work done through computer use.'
matched fp:4c06e941e0d7889b, llm:Topics: ai-agents, intelligence-layer, knowledge-base, llm, on-device-ai, personal-ai, proactive-intelligence, second-brain; description: 'An on-device LLM understands your entire life, then proactively offers to get your work done through computer use.'
matched fp:4c06e941e0d7889b, llm:Topics: ai-agents, intelligence-layer, knowledge-base, llm, on-device-ai, personal-ai, proactive-intelligence, second-brain; description: 'An on-device LLM understands your entire life, then proactively offers to get your work done through computer use.'