Letterbook Docs
Organize internal knowledge
Structure company policies and product notes so support answers stay consistent.
Last updated June 25, 2026
What this guide covers#
Internal knowledge is support guidance for your team and AI agent. It can include policies, standard operating procedures, escalation rules, product limitations, and decision trees that should not necessarily be published to customers.
Use this guide to keep internal knowledge usable as more articles are added.
Organization steps#
- Group articles by support area, such as billing, account access, troubleshooting, integrations, and product limits.
- Mark whether an article is customer-facing or internal-only.
- Give each high-risk article an owner.
- Retire outdated policy notes instead of leaving conflicting guidance in place.
- Review articles after product, pricing, legal, or billing changes.
- Test updated articles against real or realistic conversations.
What belongs in internal knowledge#
Add internal knowledge for guidance that affects support decisions:
- Refund and cancellation policy
- Billing exception rules
- Identity verification steps
- Account access and security escalation paths
- How to interpret database, Stripe, Shopify, Google Play, or other integration fields
- Which actions require approval
- Known product limitations and workarounds
- Tone and communication preferences
Do not store secrets, raw credentials, private keys, or customer data that belongs in a connected system of record.
Public versus internal content#
Use public articles for customer-facing explanations. Use internal knowledge for decision-making guidance.
For example, a public article might explain how cancellation works. An internal article should explain when to approve a refund, what data to check, and when to escalate.
For the full visibility and AI-editing distinction, see Internal vs external knowledge.
Maintenance cadence#
Use a faster review schedule for high-risk areas:
| Knowledge area | Suggested review |
|---|---|
| Billing, refunds, and cancellations | After every pricing or policy change |
| Security, privacy, and account access | At least monthly, and after any incident |
| Product troubleshooting | After relevant releases |
| Integrations and data fields | After schema or provider changes |
| Tone and escalation guidance | After support QA reviews |
Signs an article needs work#
Update an article when:
- Teammates keep correcting the same AI mistake.
- Customers ask the same follow-up after receiving a reply.
- A policy has changed but old decisions still appear in drafts.
- The article describes what to say but not what data to check.
- Two articles give conflicting guidance.