AI Playbook for Observed Collection Authorization
A DOT collector is directed to conduct a directly observed collection for a pipeline employee on a return-to-duty test. The employee objects, claiming the observation requirement was not properly authorized. The collector has the DER's written directive but the employee is refusing to cooperate.
When to use this playbook
- Use this playbook when the decision looks like the situation above: A DOT collector is directed to conduct a directly observed collection for a pipeline employee on a return-to-duty test.
- It is a fit when you have source files in hand and need a structured, reviewable analysis — not a generic chat answer about "Observed Collection Authorization".
- Do not use it as a substitute for licensed, legal, clinical, or authorized official judgment in the domain.
What you'll need
- DER's written directive for observed collection
- DOT 49 CFR Part 40 observed collection authorization requirements
- Employee's written objection
- DOT PHMSA return-to-duty requirements
- Prior SAP evaluation and return-to-duty test authorization
Attachments: Documents (Documents)
The Prompt
You are a Medical Review Officer and compliance officer advising on an observed collection dispute at the collection site. I am attaching: Work only from the attached source files. If a conclusion is not supported, say so. Produce: 1. Verify the legal basis for the observed collection: does return-to-duty qualify under DOT 49 CFR Part 40, and was the authorization properly issued? 2. Assess the DER's written directive: does it include all required elements under 49 CFR Part 40.67? 3. Determine the consequence if the employee continues to refuse: is this a refusal to test, and what is the regulatory outcome? 4. Assess the employee's objection: is there a legitimate regulatory basis for the objection? 5. Tell the collector the exact steps to take in the next 30 minutes and what to document if the employee leaves the site. Call out where independent models are likely to disagree, and list follow-up documents a reviewer should request.
What to expect
- Observed collection legal basis verification
- DER directive adequacy assessment
- Refusal-to-test consequence analysis
- Employee objection merit assessment
- Collector action sequence and documentation instructions
Review before you act
- Validate this output against source files before relying on it: Verify the legal basis for the observed collection: does return-to-duty qualify under DOT 49 CFR Part 40, and was the authorization properly issued?.
- Validate this output against source files before relying on it: Assess the DER's written directive: does it include all required elements under 49 CFR Part 40.67?.
- Validate this output against source files before relying on it: Determine the consequence if the employee continues to refuse: is this a refusal to test, and what is the regulatory outcome?.
- Validate this output against source files before relying on it: Assess the employee's objection: is there a legitimate regulatory basis for the objection?.
- Confirm every cited figure, date, counterparty, or requirement against the attached originals — models compress and can drop a qualifier.
- Treat disagreement between models as a review item, especially on classification, materiality, and recommended next action.
- Do not authorize an operational, clinical, legal, credit, or enforcement action solely because the models agree.
Why compare models on this
For Observed Collection Authorization, running the same attachments across independent models is useful because the hard part is classification and completeness, not fluency. The workflow is already designed to surface observed collection legal basis verification; der directive adequacy assessment; refusal-to-test consequence analysis; employee objection merit assessment. Those are comparison artifacts — they only exist if more than one model runs. Models disagree on whether an irregularity is fatal to custody, whether a prescription explains a result, and whether observation is authorized. Those splits are MRO work, not auto-verification.
See governed multi-model AI on your own prompt
Compare GPT-5, Claude, and Gemini side by side, with human review and a decision record built in.

