Research workflows on the AI Agents track. Research agents gather sources, compare them, and keep a bibliography. They are not allowed to invent a citation. Every claim that matters links to a fetched source.
This lesson assumes you already worked through Agents that edit code.
The idea in practice
Tools: search, open_url, save_note. Notes store url, quote, and the claim it supports. The final brief lists unresolved questions.
A concrete check
goal = {
'track': 'AI Agents',
'lesson': 'Research workflows',
}
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
A fluent brief with no saved notes is unreviewable. 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
Define the note schema and one rule that rejects a claim with no url.
Self-check
- Say Research workflows 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.