whatever I think might be useful later, I capture through the web clipper extension. [0]
In the meanwhile, wikipedia ships wikidata, which uses RDF dumps (and probably 8x less compressed than it should be).
https://www.wikidata.org/wiki/Wikidata:Database_download
There is room for a third option leveraging commercial columnar database research.
One "popular" example for those whose horizon doesn't extend over US country borders:
"Hurricane Katrina devastated communications infrastructure across the Gulf Coast, incapacitating telephone service, police and fire dispatch centers, and emergency radio systems. Almost three million customer phone lines were knocked out, telephone switching centers were seriously damaged, and 1,477 cell towers were incapacitated. Most of the radio stations and many television stations in the New Orleans area were knocked off the air. Paul McHale, the Assistant Secretary of Defense for Homeland Defense, summarized the damage by stating, “The magnitude of the storm was such that the local communications system wasn’t simply degraded; it was, at least for a period of time, destroyed."
https://georgewbush-whitehouse.archives.gov/reports/katrina-...
"Our preparedness culture must also emphasize the importance of citizen and community preparedness. […] Thus, citizens and communities can help themselves by becoming more prepared. If every family maintained the resources to live in their homes without electricity and running water for three days, we could allocate more Federal, State, and local response resources to saving lives. Similarly, if every family developed their own emergency preparedness plan, they almost certainly would reduce the demand for outside emergency resources. As the 9/11 Commission Report states, “One clear lesson of September 11 is that individual civilians need to take responsibility for maximizing the probability that they will survive, should disaster strike."
https://georgewbush-whitehouse.archives.gov/reports/katrina-...
Offline access and local models aren’t about assuming collapse—they’re about treating knowledge as infrastructure instead of something implicitly guaranteed.
That feels more like resilience than pessimism.
AlexNet -> Tansformers -> ChatGPT -> Claude Code -> Small LMs serving KBs
Large LLMs could have a role in efficiently producing such KBs.
This isn’t prepping for anything it’s cosplaying as a vault dweller.
P.S. Having TED talks as part of the “educational” curriculum of this project is probably the biggest circle jerk imaginable.

100% Free & Open Source
Wikipedia, AI, maps, and education tools running on your own hardware — completely free. No internet required.
Get Project NOMAD Free on GitHubSee What's Inside

Node for Offline Media, Archives, and Data — a free, open source offline server you install on any computer. Download the content you want, and it works without internet — forever. Similar products cost hundreds of dollars. Project NOMAD is free.
Wikipedia, guides, medical references
Run LLMs completely offline
Navigate without cell service
Khan Academy & courses offline
Whether you're planning for emergencies or living off-grid, Project NOMAD has you covered.
When infrastructure fails, NOMAD keeps working. Medical references, survival guides, and encyclopedic knowledge — no internet required.
Cabin, RV, or sailboat — bring a complete library, AI assistant, and offline maps wherever you go. True digital independence.
Run local LLMs, self-host your knowledge base, own your data. Built for beefy hardware and those who want full control.
Khan Academy, Wikipedia for Schools, and more — complete learning resources for families anywhere, even without connectivity.
Project NOMAD bundles best-in-class open source tools into one cohesive system.

Powered by Kiwix
Offline Wikipedia, Project Gutenberg, medical references, repair guides, and more — terabytes of human knowledge at your fingertips.

Powered by Ollama
Run powerful large language models completely offline. Chat, write, analyze, code — all without sending data anywhere.

Powered by OpenStreetMap
Full offline mapping with OpenStreetMap data. Navigate, plan routes, and explore terrain without any cell service.

Powered by Kolibri
Khan Academy courses, educational videos, interactive lessons — complete K-12 curriculum available offline.
Watch the full walkthrough to see what Project NOMAD can do on your hardware.
Other offline products charge hundreds and lock you into specific hardware. Project NOMAD runs on any PC you choose — with GPU-accelerated AI — for free.
NOMAD runs on serious hardware so you get serious AI. While other offline products are stuck on Raspberry Pis, NOMAD supports GPU-accelerated inference on models with real intelligence. Community builds range from refurbished desktops to GPU-powered rigs scoring 10 to 95 on the NOMAD Benchmark.
Looking for something lightweight for a Raspberry Pi? Check out projects like Internet in a Box. Project NOMAD is for when you want the full experience.
View Hardware GuideSee the Leaderboard
ProcessorAMD Ryzen 7 with Radeon graphics or Intel i7+
Memory32 GB RAM
GraphicsIntegrated AMD Radeon 780M+, or dedicated NVIDIA GPU
Storage1 TB SSD
Ubuntu or Debian-based Linux recommended. Windows support via Docker Desktop for development.

One command on any Ubuntu or Debian machine. Project NOMAD handles the rest.
$ curl -fsSL https://raw.githubusercontent.com/Crosstalk-Solutions/project-nomad/main/install/install\_nomad.sh -o install_nomad.sh && sudo bash install_nomad.sh
Requires Ubuntu 22.04+ or Debian 12+. Docker is installed automatically if needed.
Project NOMAD is free, open source, and ready to deploy. Get started in minutes with a single command.
Get Project NOMAD Free on GitHub
or
New content packs, AI model recommendations, and hardware guides. No spam, unsubscribe anytime.
No subscriptions, no paywalls. Project NOMAD is funded by the community that uses it.