Nino ChavezSearch site
Learning path8–12 weeks5 stages

Guided path

Builder

Build production apps with AI

Move from an AI-assisted prototype to a working system with specifications, review points, and operating evidence.

Start when
You need to turn an AI-assisted prototype into a production application.
What you’ll make
Deployed production application
3 examples

See the work behind the path

Each example shows work, a session, or writing that shaped this path. Use it to decide whether the instruction fits your work.

WorkRally HQA live product and event operation built and run as one system.SessionThe Browser Is a Shell CommandA practical example of giving an agent a repeatable inspection tool.WritingThe Scaffolding the Agent Doesn’t BuildWhy generated output still needs human judgment and operating structure.
Fit check

Know whether this is your path

Use this path if

  • You are a developer who learns by building.
  • You want production behavior, not another isolated demo.
  • You are comfortable with JavaScript, TypeScript, or web applications.

Choose another path if

  • You want a passive curriculum.
  • You need a certificate rather than a working artifact.
  • You are not ready to review generated code and product decisions.
Working concept

The application is the receipt

A credible build path ends in deployed behavior another person can inspect. Prompts, screenshots, and passing checks support that claim; none replaces the running product.

Path / 00–04

Five stages to the finished work

Each stage has a practical stopping condition. Move on when the checkpoint is true, not when the time estimate expires.

  1. 00

    Orientation

    1–2 days
    Goal

    Learn the roles models, retrieval, and agents can play.

    Done when

    You can explain the difference between a model, retrieval, and an agent workflow.

  2. 01

    Ship your first AI app

    1 week
    Goal

    Build and deploy one narrow end-to-end behavior.

    Done when

    Another person can open the application and complete its main action.

  3. 02

    Add specs, data, and workflows

    2–4 weeks
    Goal

    Make product intent and state durable outside the model.

    Done when

    The application has a specification, persistent data, and an explicit access boundary.

  4. 03

    Use multiple models and tools

    4–8 weeks
    Goal

    Move from a chatbot response to a bounded agent workflow.

    Done when

    The system can choose and use a tool without gaining unbounded authority.

  5. 04

    Operate the production system

    8+ weeks
    Goal

    Measure cost, behavior, failures, and output quality.

    Done when

    You can inspect runtime metrics and validate important outputs at their source.