AI Campus • Practical resources
Autonomous AI Agents & Tool Calling
Make tool use bounded, observable and recoverable before increasing autonomy.
Before you start
- Python functions, JSON and basic error handling. A local deterministic state machine is enough to begin.
The reading and planning exercises need no paid course or API. Optional inference, training, IDE features and hosted services may require hardware, accounts or paid access. Use the local or manual alternative before spending money.
Start with the official learning material
Free-to-read references checked 3 November 2026. Reading documentation does not activate a provider or create a subscription.
Practice with a concrete exercise
1. Represent state explicitly
Read the overview. Sketch plan, retrieve and review states for a three-source research question. Define the state fields and termination condition before adding a model.
2. Validate a read-only tool
Create a function that reads only named local source files. Reject unknown filenames and enforce a maximum file size. Return a structured success/error value rather than throwing away failures.
3. Test the stop rules
Set a five-step limit and a one-retry budget. Simulate a missing source, an invalid tool argument and an instruction embedded inside source text. Verify the controller stops and does not treat untrusted text as a new command.
Check what you learned
- Every tool call has an input and recorded result.
- The loop stops at its budget even if the goal is incomplete.
- Any action outside the permitted tool contract is rejected.
Keep your artifact, observed checks and remaining limitations in a short README. Mark any exercise you only designed rather than executed.