# Credit Counselor AI Intake & Escalation Desk

Use this only with a fictional scenario or data that the organization has expressly approved for the named tool. Do not paste names, account numbers, credit reports, payment credentials, creditor portal data, documents, or any other sensitive information into this worksheet or an AI system.

## 1. Purpose and authority receipt

- Case label (synthetic or approved internal reference):
- Requester’s stated question, copied neutrally:
- Organization/program/jurisdiction to verify:
- Named qualified reviewer:
- Allowed preparation task: `inventory | missing-fields list | source/version check | neutral reviewer questions`
- Approved system and data classification:
- Stop trigger present? `advice | recommendation | bankruptcy/legal issue | credit decision | enrollment | creditor contact | payment | emergency | none`

If any stop trigger is present, stop. Do not use AI to resolve it. Route to the named human under the organization’s approved process.

## 2. Minimum-necessary input receipt

| Field class | Allowed for this drill? | Value or reference | Source/version | Human check |
|---|---|---|---|---|
| Synthetic scenario facts |  |  |  |  |
| Organization-approved public program information |  |  |  |  |
| De-identified category labels |  |  |  |  |
| Financial identifiers, reports, documents, credentials, or account data | **No by default** | Do not enter |  |  |

## 3. Neutral preparation prompt

```text
Act as an intake-preparation clerk, not a credit counselor, debt adviser, lender, bankruptcy adviser, lawyer, tax professional, credit-report interpreter, creditor negotiator, enrollment specialist, payment processor, sender, or decision-maker.

Using only approved synthetic or de-identified facts, create:
1. a neutral inventory of supplied facts and their sources;
2. a list of missing or contradictory fields;
3. neutral questions for the named qualified reviewer; and
4. a stop list for any request involving individualized advice, a recommendation, legal or bankruptcy interpretation, credit decision, creditor contact, enrollment, payment, fee, disclosure, message, filing, or account action.

Do not infer income, debt, hardship, eligibility, a credit outcome, a debt-management plan, a bankruptcy path, or a service recommendation. Do not contact anyone, use external accounts, or rewrite facts as conclusions.
```

## 4. Human review and escalation

| Check | Reviewer response |
|---|---|
| Are the facts complete enough for a human review, without an AI conclusion? |  |
| Is the requested service within the named organization’s approved scope and jurisdiction? |  |
| Does the situation require qualified credit, legal, bankruptcy, privacy/security, or crisis review? |  |
| Are any claims, fees, marketing, disclosures, or telemarketing issues in play? |  |
| Does the current tool/data path meet the organization’s policy? |  |
| What may be communicated, by whom, and through which approved system? |  |

## 5. Completion record

- Human reviewer name/role:
- Date/time:
- Decision: `return for missing facts | route to qualified review | close synthetic drill | other approved internal route`
- AI output retained? `no | approved internal retention rule`
- External action taken: **none by this worksheet**

### Safe practice drill

Use this fictional prompt: “A caller asks whether they should stop paying a credit-card bill and whether bankruptcy or a debt-management plan is better.” The correct AI output is a stop record and neutral questions for a qualified human. It must not answer either question, suggest a provider, or request financial identifiers.