From a repeated task to a tested AI workflow.
A free AI Automation course: 20 modules, 60 lessons, 20 guided workflow practicals, 120 quiz questions, and six projects with downloadable notebooks.
Start the first lessonBefore you begin: prerequisites, tools, and learning plan
For students, builders, and small teams. Start with basic Python variables, dictionaries, lists, functions, and loops. Use the Python course foundations if these are new to you. No AI or automation experience is assumed.
Each module has two lessons and one guided practical. Run a worked example, implement a function, pass input and failure checks, and answer two quiz questions. Plan 3 to 5 hours each week, plus time for six larger projects.
Required exercises use standard Python in the browser with fictional records and simulated AI responses. No paid account, API key, or external automation service is required. The first run downloads Python. Notebooks also work locally. An optional visual workflow plan can be mapped to a tool you already use, such as n8n or Make.
The lab does not call a live AI model, send messages, or connect your accounts. Simulations teach workflow logic, not measured model quality. Before a live pilot, add provider-specific authentication, durable state, evaluation, access controls, and approval for external actions.
Review Python foundationsYour course outline
The interactive lessons need JavaScript. The full outline and module readings are also available as plain pages.
- Find the right work to automate
Map a repeated task, choose a useful goal, and decide where AI belongs.
- Work with records and decisions
Use small Python functions to clean inputs and route work.
- Exchange and validate JSON
Read structured messages and reject invalid input before it spreads.
- Handle triggers and duplicate events
Filter event types and prevent repeat deliveries from creating repeat work.
- Understand APIs and HTTP
Represent requests and interpret responses without making live calls.
- Write prompts for repeatable tasks
Define the task, separate source data, and make missing information visible.
- Validate AI output
Treat generated labels and JSON as proposals that must pass checks.
- Build an AI-assisted triage flow
Compare a simple baseline with simulated predictions and review uncertain work.
- Extract facts with evidence
Preserve missing values and connect extracted fields to source text.
- Ground answers in approved sources
Retrieve relevant material, respect access, and abstain when support is missing.
- Control tool use and agent loops
Validate tool proposals and bound the work an AI agent can request.
- Add meaningful human approval
Tie approval to the exact draft and keep changed or expired work on hold.
- Recover from failures
Use bounded retries and separate temporary failures from permanent problems.
- Save state and resume work
Track durable outcomes and design replay-safe steps.
- Schedule and batch work
Use explicit time zones and prepare bounded batches and useful digests.
- Evaluate the whole workflow
Measure errors, review load, and regressions with fixed examples.
- Track cost and operational health
Set budgets, record useful logs, and notice when a workflow degrades.
- Protect data and action boundaries
Treat external text as data and keep sensitive fields out of unnecessary steps.
- Move from a lab to a small pilot
Map the workflow into visual tools and prepare a controlled release.
- Deliver an automation capstone
Combine the course ideas into a tested workflow and a clear handover.