Token and tool cost on the AI Agents track. Agents spend tokens on every step and money on some tools. A loop that re-sends the whole history grows cost linearly with steps, then worse. Budgets are part of the design.
This lesson assumes you already worked through Tool errors and retries.
The idea in practice
Set a max token budget and a max tool-call budget per run. Summarize older steps instead of resending raw pages. Cache repeated reads.
A concrete check
goal = {
'track': 'AI Agents',
'lesson': 'Token and tool cost',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Pasting full web pages into the next prompt on every step will dominate the bill and crowd out the goal. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Estimate tokens for a 10-step run if each observation is 2,000 tokens and you resend all of them. Then estimate if you keep only a 300-token summary.
Self-check
- Say Token and tool cost in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.