Build a Production AI Agent in 6 Weeks

Go deeper · Build AI agents · See all steps

API calls are the easy part. Architecture is the differentiator.

Anyone can get an agent to work once. The hard part is the system around it: validation, boundaries, fallbacks, and the structure that makes it predictable instead of flaky. In six weeks you build Expense AI Agent and ship it, with two senior engineers reviewing your pull request every week.

Juan José Expósito González Bob Belderbos Coached by Juanjo & Bob · meet your coaches ↓
Book a call to reserve your seat → Full curriculum →

What you ship in 6 weeks

150+
passing tests
95%+ coverage · pytest exits 0
3
working interfaces
CLI · Telegram · Web (REST + dashboard)
1
Docker deployment
multi-stage build · CI/CD · ready to ship

A web dashboard to classify, review, and analyze expenses, with a REST API behind it:

Dashboard with expense analytics and charts AI-powered expense classification form Expense list with categories and pagination

And the same agent on the go, as a Telegram bot that confirms before it acts:

Telegram bot classifying a book purchase as Education Telegram bot parsing a multilingual metro expense Telegram bot summing a multi-item cinema expense

How it works: test-driven coaching

  1. Red.Each week you get a complete pytest suite, all failing.
  2. Write.You write the production code that makes it pass.
  3. Green."Done" is objective: pytest exits 0. That's the baseline.
  4. Senior review.We review your PR line by line: naming, service boundaries, and why this pattern over another.

A product, not a notebook

Function calling, structured outputs, a Telegram bot, a web dashboard, and a Docker deploy on your GitHub.

Architecture that transfers

Protocols, the repository pattern, service layers, dependency injection. How senior engineers build any system.

Judgment over rubber-stamping

The instinct to tell whether an agent's architecture holds, instead of accepting whatever the model produces.

Agentic AI program overview

Code review that levels you up

Real threads from cohort 1, anonymized. Click a screenshot to zoom in.

Review thread: token prices are per 1M tokens, so the cost estimate is off by a million
A units bug a mocked test won't catch: prices per 1M tokens, used per token
Review thread: the student finds that the provided tests contradict each other, and the coaches update them
The developer found that our own tests contradicted each other, so we changed the tests
Review thread: the student corrects the suggested OpenAI type and ty confirms it
It goes both ways: the developer corrects our suggested type, and ty confirms it
Review thread: a config suggestion the student turns into centralized settings
A week 2 suggestion the developer turned into centralized config
Review thread: config moved to typed pydantic-settings with a test for the missing key
Closing the loop: typed pydantic-settings, lazy reads, and a test for the missing key
Review thread: a test that passes by accident because Exception stringifies its arguments
"Why is the test passing? Maybe that's good enough." A green test isn't always a correct one
"It's different having done it than even reading about it. If somebody had written an explanation, a week later I'd have a vague idea. This, I think I understand."
— Jeff H., built an agentic app in 6 weeks

Join the cohort

€2,000 one-time · 6 weeks · ~10 hrs/week

  • Weekly line-by-line PR review from two senior engineers
  • 150+ pre-written tests: the tests are the assignment, your code is the answer
  • A full-stack agent you deploy: CLI, Telegram bot, FastAPI + dashboard, Docker, CI/CD
  • Weekly live call with both coaches, recorded, plus async support all week
  • Capped at 6 developers, with a private cohort channel in our Slack community during and after
Book a call to reserve your seat →

Your coaches

Juan José Expósito González

Juan José Expósito González: Python and AI mentor, PhD engineer. Guides developers from Python basics to production AI, with a focus on clean, testable, deployable architecture.

Bob Belderbos

Bob Belderbos: developer coach and builder, co-founder of Pybites, 150+ developers coached. Bob brings the architecture perspective: clean code, testable design, and getting AI prototypes production-ready.

Next on the path: the Rust cohort, or 1:1 coaching on your own AI project.