WORKING MODELS // LOCAL-FIRST AI operational · build 1.0
wm Working Models
Get early access
[ practical ai · local-first ]

Make AI models
do the work.

Small AI systems that do real jobs — local, affordable, and yours to keep. Tools you can run today, plus a few ways to go deeper.

Browse what's coming watch on youtube ↗
71k on youtube/weekly field notes/your own machine
how it works running locally
Your task
14 emails in
Working Models
local model
Done
12 handled

Give it a job. A local model does the work.
You get the result back — on your own machine.

01 / WHAT YOU CAN RUN

Tools that put a model to work.

Buy them, or grab the free ones. Each does one real job — no subscriptions to babysit, runs on your own hardware.

PKG-01 · SOON
Local Setup Pack

Drop-in configs to run a capable model on your own Mac — context, memory and prompts dialed in.

get notified →
APP-01 · FREE
Inbox Agent

Hand your support queue to a local model. It drafts replies, resolves the easy ones, flags the rest.

get notified →
EDU-01 · SOON
Build Your First AI Employee

The full system, start to finish — pick a job, wire up a local model, and let it run.

join the waitlist →
02 / GO DEEPER
Course & community

Learn the whole approach and build alongside others doing the same. Opening once there's a crowd to fill it.

join the early list →
03 / WORK WITH ME
For teams adopting AI

Consulting, talks and webinars on putting models to work — selectively, when the fit is right.

Start a conversation →
04 / PROOF

Seen the work? 71,000+ on YouTube.

youtube ↗ blog ↗
Gemma 4 12B on a 16GB Mac Mini Is Surprisingly Capable
watch on youtube ↗
Hermes Agent: Build Your First AI Agent (Gemma 4)
watch on youtube ↗
Claude Code + LM Studio: FREE Unlimited AI Agents
watch on youtube ↗
[ early access ]

Get first access when it ships.

Setup packs, the Inbox Agent and the course are on the way. Subscribers get the early list — and the weekly field notes in between.

Get early access →

no spam · unsubscribe anytime