🚨 Andrej Karpathy thinks RAG is broken. He published the replacement 2 days ago. 5,000 stars in 48 hours.
It's called LLM Wiki.
A pattern where your AI doesn't retrieve information from scratch every time. It builds and maintains a persistent, compounding knowledge base. Automatically.
RAG re-discovers knowledge on every question. LLM Wiki compiles it once and keeps it current.
Here's the difference:
RAG: You ask a question. AI searches your documents. Finds fragments. Pieces them together. Forgets everything. Starts over next time.
LLM Wiki: You add a source. AI reads it, extracts key information, updates entity pages, revises topic summaries, flags contradictions, strengthens the synthesis. The knowledge compounds. Every source makes the wiki smarter. Permanently.
Here's how it works:
→ Drop a source into your raw collection. Article, paper, transcript, notes.
→ AI reads it, writes a summary, updates the index
→ Updates every relevant entity and concept page across the wiki
→ One source can touch 10 to 15 wiki pages simultaneously
→ Cross-references are built automatically
→ Contradictions between sources get flagged
→ Ask questions against the wiki. Good answers get filed back as new pages.
→ Your explorations compound in the knowledge base. Nothing disappears into chat history.
Here's the wildest part:
Karpathy's use case examples:
→ Personal: track goals, health, psychology. File journal entries and articles. Build a structured picture of yourself over time.
→ Research: read papers for months. Build a comprehensive wiki with an evolving thesis.
→ Reading a book: build a fan wiki as you read. Characters, themes, plot threads. All cross-referenced.
→ Business: feed it Slack threads, meeting transcripts, customer calls. The wiki stays current because the AI does the maintenance nobody wants to do.
Think of it like this: Obsidian is the IDE. The LLM is the programmer. The wiki is the codebase. You never write the wiki yourself. You source, explore, and ask questions. The AI does all the grunt work.
NotebookLM, ChatGPT file uploads, and most RAG systems re-derive knowledge on every query. This compiles it once and builds on it forever.
5,000+ stars. 1,294 forks. Published by Andrej Karpathy. 2 days ago.
100% Open Source.
Mission success! Alpha Flight 7 achieved nominal performance and validated key systems ahead of our Block II configuration upgrade. This test flight also delivered a demonstrator payload for @LockheedMartin . Congratulations to the entire Stairway to Seven team! Read more: https://t.co/mz3wUbc8Hi
Stepping into the Stairway to Seven! The team completed the launch readiness review and we are GO to move into final operations for Alpha Flight 7.
The two-hour launch window for this test flight will open at 5:50 pm PT tomorrow. Our livestream with @NASASpaceflight will begin 20 minutes before liftoff. https://t.co/8UsuO9IGaq
A year ago, Firefly ignited the lunar economy as the first commercial company to successfully land on the Moon. Congrats to all our Fireflies who made their mark on the lunar surface and thank you @NASA for trusting us along the ride!
Alpha is vertical on the pad and getting ready for our static fire ahead of the Stairway to Seven mission! Check out our merch store and pick up a Flight 7 t-shirt and mission patch. https://t.co/j2nxNCAiUo
#BlueGhost got her first diamond ring! Captured at our landing site in the Moon’s Mare Crisium around 3:30 am CDT, the photo shows the sun about to emerge from totality behind Earth. Hope to have more shots to share soon! #BGM1
Shortly after landing, #BlueGhost deployed four tethered Lunar Magnetotelluric Sounder (LMS) electrodes to the surface and a 8-foot mast above our top deck. Look closely in the video below! These instruments will allow @NASA and the @SwRI team to study the deep interior of the Moon - up to 2/3rds the distance to the Moon's center - to learn more about the structure and composition of the Moon’s mantle.
Watch Firefly land on the Moon! After identifying surface hazards and selecting a safe landing site, #BlueGhost landed directly over the target in Mare Crisium. A historic moment on March 2 we'll never forget. We have Moon dust on our boots! #BGM1
At this point in our roadtrip, we're 12 hours away from parking our lunar lander on the Moon! Blue Ghost is getting ready to begin a series of descent operations to prepare for a soft touchdown. During these final hours of travel, we'll provide live updates here and on our #BGM1 livestream with @NASA, beginning at 1:20 am CST: https://t.co/BQapBq0HmL
Looking at the Moon in a new light as Blue Ghost lands tonight! Livestream with @NASA begins at 1:20 am CST. During landing, Blue Ghost's bandwidth will be dedicated to critical descent operations, but you'll see live data from Mission Control in Texas! https://t.co/KCYPAogMJ2
Our #GhostRiders completed another lunar orbit maneuver with a 3 minute, 18 second burn early this morning. This maneuver moved the lander from a high elliptical orbit to a much lower elliptical orbit around the Moon. Shortly after the burn, Blue Ghost captured incredible footage of the Moon's far side, about 120 km above the surface.
In this orbit, the team will experience planned rolling comms blackouts as Blue Ghost goes around the far side of the Moon. When on the near side, the team will continue to downlink data and finalize the plan for our next maneuver that will get Blue Ghost even closer to the lunar surface and keep us right on track for landing on March 2. #BGM1
Yesterday I got to visit this test stand, no engine firing though, except for the heater blowing hot air on the technicians working there. Hot enough to melt clothes….
Ran into @seanparkerphoto at the world famous @ocairnssuites . He snagged a beautiful photo of our last launch and we snagged him as an official photographer for Firefly. Fingers crossed for clear skies tomorrow!