AI Playbook for State-Federal Grant Compliance
A state agency administering a $48M federal grant has received a compliance letter identifying 3 issues: unallowable costs of $620,000, subrecipient monitoring failures, and a procurement that did not follow uniform guidance. The state must respond in 45 days.
When to use this playbook
- Use this playbook when the decision looks like the situation above: A state agency administering a $48M federal grant has received a compliance letter identifying 3 issues: unallowable costs of $620,000, subrecipient monitoring failures, and a procurement that did not follow uniform guidance.
- It is a fit when you have source files in hand and need a structured, reviewable analysis — not a generic chat answer about "State-Federal Grant Compliance".
- Do not use it as a substitute for licensed, legal, clinical, or authorized official judgment in the domain.
What you'll need
- Federal compliance letter and finding details
- Grant agreement and approved budget
- Cost documentation for the $620,000 in questioned costs
- Subrecipient monitoring policies and documentation
- Procurement records for the flagged acquisition
- 2 CFR 200 Uniform Guidance requirements
Attachments: Documents (Documents)
The Prompt
You are a state grant compliance officer responding to a federal grant compliance letter with 3 findings. I am attaching: Work only from the attached source files. If a conclusion is not supported, say so. Produce: 1. Assess the $620,000 in questioned costs: which are clearly unallowable, which are allowable with better documentation, and which should be disputed? 2. Assess the subrecipient monitoring failure: what specific 2 CFR 200 requirements were not met and what is the passthrough liability? 3. Assess the procurement finding: material deviation from uniform guidance or a technical deficiency? 4. Develop the corrective action plan for each finding. 5. Tell me which questioned costs the federal agency is most likely to demand repayment on and what the appeal process is. Call out where independent models are likely to disagree, and list follow-up documents a reviewer should request.
What to expect
- Questioned cost assessment (unallowable vs. allowable vs. disputable)
- Subrecipient monitoring liability analysis
- Procurement finding severity assessment
- Corrective action plan
- Debt repayment risk and appeal process
Review before you act
- Validate this output against source files before relying on it: Assess the $620,000 in questioned costs: which are clearly unallowable, which are allowable with better documentation, and which should be disputed?.
- Validate this output against source files before relying on it: Assess the subrecipient monitoring failure: what specific 2 CFR 200 requirements were not met and what is the passthrough liability?.
- Validate this output against source files before relying on it: Assess the procurement finding: material deviation from uniform guidance or a technical deficiency?.
- Validate this output against source files before relying on it: Develop the corrective action plan for each finding.
- 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 State-Federal Grant Compliance, 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 questioned cost assessment (unallowable vs. allowable vs. disputable); subrecipient monitoring liability analysis; procurement finding severity assessment; corrective action plan. Those are comparison artifacts — they only exist if more than one model runs. Models disagree on exemption applicability, IG finding risk, and how aggressive a budget narrative can be. Those disagreements belong with counsel and the authorizing official.
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.

