Loading
10 weeks. First principles. Real systems.
Not a framework tutorial. A cohort deep-dive that teaches you to look at any agent system — LangGraph, CrewAI, your own code — and immediately understand what's underneath.
Four principles that shape every session of this program — not as marketing copy, but as actual constraints on how the curriculum is run.
Frameworks come after the pattern — framed as "here's how production tooling implements what you already built," not as the thing itself.
Papers, engineering blogs, and source code are the default reading — not third-party explainer videos.
Deep work happens in your own time. The daily hour is for framing, dissection, and debugging together — not lecturing.
No week passes without a working artifact. By Week 10 you leave with 9 real shipped projects.
Developers comfortable with a general-purpose language and working with APIs, who are new to agentic AI specifically. Some prior exposure to LLMs and prompting is assumed. A short Week 0 primer can be added for cohorts that need it — without disrupting the 10-week core.
Four acts, each building on the last. By the summit you will have built the entire agent stack from scratch — loop, memory, planning, tools, multi-agent, framework, eval, production.
"Everything else in the course is an elaboration on this loop."
"Most 'agent got dumber' problems are memory problems, not model problems."
"The difference between an agent that reacts and one that thinks ahead."
"A model is only as capable as the actions it can take."
"One agent juggling everything in a single context window stops working."
"The payoff week — everything built by hand now maps onto real production tooling."
"'It worked when I tried it' is not engineering."
"Not every problem needs an agent — and not every agent needs to be a demo."
"Opens the door to what's next, and makes AI-assisted development deliberate rather than incidental."
"Where every concept converges into one system you designed, built, and can defend."
Cohorts are small, selective, and run twice a year. If this resonates, don't wait for the next one.
Small enough that everyone is known, every question gets a real answer, and the group actually gels.
Live-only format. The value is in the room — real debugging together, real-time thinking, real peer pressure to ship.
We review every application. The cohort quality matters — each person affects the environment for everyone else.
We'll contact you as soon as the next cohort opens. Keep an eye on your inbox.