This is your beginning. A step‑by‑step path that takes you from zero to a fully working local AI agent — running on your machine, using your tools, and acting on your behalf.
Begin the Journey →Local‑first AI unlocks a new way of working with intelligent systems — private, fast, offline, and fully under your control. This Starter Path guides you through that journey in three clear stages: a beginner‑friendly introduction, a hands‑on intermediate track, and a deep advanced path for building autonomous, multi‑agent systems.
Each tier builds on the last, helping you move from curiosity to capability, and finally to mastery of agent‑native development.
The simplest way to understand local AI is to experience it. With Jan, a free and open‑source desktop client, you can run your first model instantly — no setup, no terminal, no configuration. It’s the perfect starting point for anyone curious about how local models feel in practice.
A polished desktop app for Windows, macOS, and Linux.
Jan recommends models that match your hardware.
Everything runs locally — no cloud, no accounts, no API keys.
Once you’ve experienced local AI, the next step is learning how to build with it. This intermediate path teaches you how to install models, benchmark performance, set up your development environment, and create your first working agent — a foundation you’ll build on in the advanced path.
These steps are hands‑on but approachable, giving you your first real taste of agent development.
Set up the engine that runs local models.
Choose the right models for your machine.
Measure speed, latency, and performance.
Prepare your environment for agent development.
Create a simple agent script that thinks and responds.
The advanced path takes you from “I built an agent” to “I built intelligence.” You’ll learn how to run your agent as a local service, extend it with tools, design autonomous loops, and orchestrate multiple agents working together.
This is where agent‑native development becomes powerful — and where your system begins to think, plan, act, and reflect.
Expose your agent through a local API.
Give your agent the ability to read, write, search, and act.
Teach your agent to observe, plan, act, and reflect.
Coordinate multiple agents into a single intelligent pipeline.
Ready to continue?
Begin the Intermediate Path →