# Use AI to compare fault hypotheses and choose the next check

Educational worksheet; not an executable workflow.
Reference: https://netprompt.ai/learn/ai-troubleshooting-hypothesis-testing

Task / service:
Scope and owner:
Platform and version:
Observation window and time zone:

## Practical example
Illustrative symptom: a user cannot open an internal application by name. Candidate explanations include a name-resolution problem and a later connection failure. First record the actual resolver response and the subsequent application behavior rather than immediately changing DNS.
If the returned address matches the expected service, continue testing the remaining hypotheses. If the response is unexpected, inspect the relevant DNS context. Neither branch justifies declaring a root cause without the supporting observations.

## Review checklist
- [ ] Define the symptom and expected behavior.
- [ ] List a few plausible explanations.
- [ ] For each, identify supporting and contradicting evidence.
- [ ] Choose a bounded check that separates the candidates.
- [ ] Update or retire hypotheses as new observations arrive.

## Evidence review
| Claim or question | Observation / source | Time | Known gap | Next action / owner |
|---|---|---|---|---|
| | | | | |

## Outcome
Supported findings:
Unresolved hypotheses:
Required approval, if any:
Verification result:
Reviewer and date:
