Arm yourself with AI before someone else does.
The landscape of AI is changing, and AI itself is changing every day — too rapidly to keep up with all of the happenings. But with the tools below, you can arm yourself with knowledge. Just choose the right tool for the job.
AI will probably kill us all in the end, but until then, we need to learn how to use it to fight those who WILL use it against us. AI is like a firearm: if you wish to defend yourself, you need to arm yourself with the tool and the training. Don’t get caught defenseless — ignorance is not bliss.
Three ways to use AI — pick the one that matches the job
Every tool on this page sits in one of three trust levels. Convenience goes down as privacy goes up, so match the tier to how sensitive the thing you’re working on is.
Public / Big Tech
Quite good at many tasks — just be aware that they are collecting all of your exchanges and attributes about your machine and network.
Brokered / private
Better ways to reach the big tech models. For sensitive personal information you don’t want leaked to the technocrats at Google, OpenAI, Microsoft, etc.
Local / offline
Download the model and run it on your own computer or server, 100% offline. The maximum-privacy way to use AI — and easier than you think.
No tools match that.
The “Big Tech” models
Good tools, and worth knowing. Just go in with your eyes open about what you’re trading for the convenience.
Heads up Assume every prompt is retained and attached to your account and machine. Great for general questions — not for anything you wouldn’t say out loud.
Better ways to use the big tech models
Front-ends and brokers that put capable models behind a layer you control a little better.
Private AI models
Built for the conversations you’d rather keep between you and the machine.
Enoch is also available as an offline model you can download and run locally — see the Offline AI section below.
Specialty AI models
Single-purpose tools that do one job better than a general chatbot will.
Other models worth a look
Learning to use AI
Free courses and references. An hour here is worth more than another ten tools on the list.
Stuck on something? Then simply ask AI.
Seriously, it’s that easy. Don’t know the difference among all of the various ChatGPT or Anthropic models? Just ask AI and it will tell you in as much detail as you wish. If you’re not sure how big a model your hardware will run, ask AI — it knows the answer to that question, too.
Offline AI
Applications for downloading and running models locally on your own computer or server — most for free, with 100% offline privacy.
Ollama with an Open WebUI front end is a wonderful way to run all of the Ollama models available on their Models page, but it can also run many other models with the .gguf file extension type. If you have no idea how to download and use these with the bazillions of local AI models available for free, then ask an online AI and it will show you exactly how to install and configure your own offline AI. Seriously, it’s easier than you think. Enoch is an excellent one to have, for example — it’s only 7.5GB and can be run with Ollama, LM Studio, and others. It’s excellent “non-woke” AI with tons of uses: it’s been trained on over 3000 firearms, health and wellness, off grid living and many other educational and scientific resources. Just like with online AI, there are models for specific purposes such as text generation, chat, images, medical, coding, reasoning, etc.
Apps for running models locally
Ollama
Add OpenWebUI for a beautiful user interface. Runs everything on Ollama’s Models page, plus any model packaged as a
.gguf file.
Llama.cpp
The low-level engine a lot of the other apps are built on. Great when you want maximum control and speed.
LM Studio
Point-and-click desktop app for finding, downloading and chatting with local models. The easiest starting point.
text-generation-webui
(Oobabooga) — a browser-based front end with every knob exposed. For when you want to tinker.
Offline models I recommend
Reality check: your hardware decides your speed
All of these will run on a regular laptop or desktop, but CPU-only they’ll be painfully slow and may even crash your system. A decent GPU makes them very fast — although GPUs are not all the same. Watch the parameter size when you download a model.
| Your hardware | What to expect |
|---|---|
| Raspberry Pi / laptop | 7–12B parameter models run okay. Larger than that and it gets rough. |
| Larger models | You really need a GPU. |
| ~$500 GPU | Works okay. |
| ~$3,500 GPU | An order of magnitude better. |
| Multiple GPUs | Even better performance for offline AI, if you have the resources. |
Not sure how large a model your machine can run? Ask AI — it knows the answer to that question.
See it running
Enoch on a single RTX 5090
A Debian 13 server running the Enoch model locally on one NVIDIA RTX 5090 GPU.
Uncensored AI — local, private, offline
Offline AI that will answer… ANYTHING.
Example project built with AI
Proof this stuff is usable for real work, not just conversation.
Voice Commander
A dashboard for quickly launching common tasks and apps — using buttons or your voice. Full installation instructions are included so you can build it on your own machine.
Uncensored AI — local, private, offline
The kind of AI that will answer… ANYTHING. Because it runs on your hardware, not someone else’s server.
Uncensored AI
Local, private, offline AI that will answer… ANYTHING.
Don’t get caught defenseless.
Ignorance is not bliss. Pick one tool from this page and use it today — then ask it to teach you the next thing.