# Approved-Facts Knowledge Base Packet

Sample buyer-facing packet generated from Setup Your AI private product repositories.

Source: setupyour-agent-products
Data policy: sample fixture only; no live customer records, credentials, uploads, payment data, or private transcripts.
Approval policy: human approval is required before external sends, customer-system writes, booking holds, pricing exceptions, payment-affecting actions, legal answers, refunds, or policy exceptions.

# Approved-Facts Knowledge Base: Mountain View Events

Operator: Jim

## Source materials

- venue FAQ page
- tour request form
- owner policy notes

## Success metric

- Every chatbot, front-desk, and follow-up draft can cite approved facts or escalate.

## Approved facts

- Approved fact 1: Tours run Tuesday through Saturday.
- Approved fact 2: The venue handles weddings and private events up to 180 guests.
- Approved fact 3: Booking holds, availability, discounts, and contract terms require owner approval.

## Needs owner review

- Exact September Saturday availability
- Custom discount requests
- Whether outside catering is allowed for a specific event

## Common questions

- Can we tour this weekend?
- Do you have September Saturdays available?
- Can you hold a date before we sign?

## Escalation rules

- Escalate availability, booking holds, pricing exceptions, and contract questions.
- Escalate legal, refund, safety, and accessibility questions.
- Escalate any claim that does not appear in the approved facts list.

## Agent boundaries

- Use only approved facts when drafting chatbot, front-desk, or follow-up answers.
- Ask a concise follow-up question when required details are missing.
- Escalate anything marked needs_owner_review before it reaches a customer.
- Do not invent availability, prices, policies, guarantees, discounts, legal answers, or contract terms.

## Launch checklist

- Confirm each approved fact with the owner before using it in an agent.
- Convert common questions into safe answers, follow-up questions, or escalation paths.
- Run common, ambiguous, high-value, and risky questions before launch.
- Schedule the review cadence so stale facts do not keep feeding agents.

## Acceptance checks

- Every answer source is either approved or marked needs_owner_review.
- Common questions map to approved facts, missing-detail questions, or escalation rules.
- The owner can see which facts are allowed before any chatbot or agent launch.
- No output claims CRM, calendar, email, SMS, booking, payment, or website integrations are already wired.

## Review cadence

- Review the fact sheet every month and after any policy change.

## Non-goals

- No autonomous external sending in v0.
- No CRM, calendar, email, SMS, booking, payment, or website write access until explicitly approved and scoped for the client.
- No scraping private accounts, bypassing login walls, or storing customer credentials.
- No live customer data, private transcripts, uploaded files, payment records, or secrets stored in this reusable product repo.

## Pricing anchor

- Approved-facts knowledge base setup starts around $2,500.