Cohort course · 16 sessions · 4 months

In 16 weeks, ship a deployed AI agent — and build every layer yourself.

One codebase, four milestones: a chatbot that speaks, a coding agent that acts, a full harness that plans and remembers, and a live product on a public URL. Graded by eval suites, proven on Demo Day.

See the curriculum ✓ You're enrolled — see your inbox for cohort details
mouthhandsharnessworld # chatbot to deployed agent, week 1 to 16
Curriculum

Four milestones, one growing organism

00 · Twin track

Tech + Business

Every technical session pairs with a business mirror: price, position, and sell the exact agent you're building.

▲ B01–B16 mirror S01–S16
M1 · Weeks 1–4

Give the LLM a Mouth

API fundamentals, system prompts, streaming, conversation state, and your first eval suite.

▲ Ship: a chatbot with memory
M2 · Weeks 5–8

Give the LLM Hands

Tool schemas, the agent loop, file & shell tools — and the safety of letting a model act.

▲ Ship: ClawBot, a terminal coding agent
M3 · Weeks 9–12

Build the Full Harness

Permissions, hooks, subagents, context compaction, skills, memory — the mechanisms behind real coding agents.

▲ Ship: a comprehensive agent
M4 · Weeks 13–16

Ship It to the World

FastAPI service, Docker, Railway deploy, prompt-injection hardening, cost control — then Demo Day.

▲ Ship: your agent, live on a URL
Pricing

Start with a $50 application fee.

Submit the application fee to reserve your review slot. Accepted students receive cohort details, schedule, and next steps by email.

  • 48 hours live instruction — 16 sessions × 3 hours, weekly
  • 4 graded milestone builds with machine-run eval suites
  • Peer + instructor code review at every milestone
  • Demo Day in front of invited reviewers & hiring partners
  • Business twin track — pricing, ROI cases, GTM for your agent
Application fee · next cohort
$50

30 seats per cohort. Applications are reviewed on a rolling basis.

Secure checkout by Stripe · Google sign-in required · promo codes accepted at checkout
FAQ

Questions

Do I need machine learning experience?

No. We use models; we don't train them. You need intermediate Python, basic Git, and about five hours a week outside class.

What if I can't attend a live session?

Sessions are recorded for enrolled students. Milestone reviews and Demo Day are live-participation, so plan around those four dates.

What does the course cost beyond the application fee?

An LLM API key (roughly $5–15/month of usage, budget-capped in class exercises) and a free-tier Railway account for deployment. Full cohort tuition details are shared during review.

Is there a refund policy?

Full refund before session 2, prorated through week 4, none after milestone 1 grading. Email us and we'll process it through Stripe.

What exactly do I walk away with?

A public repo with four months of commits, a deployed agent on a live URL, passing eval results for four milestones, a Demo Day recording, and a business case for the product you built.