Knowledge
Governance & quality
Keep knowledge accurate, scoped, and reviewable so automation stays trustworthy.
Why governance matters
Workflows and bots can only be as correct as the knowledge they’re allowed to reference. A small amount of curation prevents:
- contradictory answers across channels
- low-trust automation (“the bot makes things up”)
- operator confusion in Shared Inbox
Recommended operating rules
- One canonical source per topic: if two items cover the same policy, pick one and archive the other.
- Use Q&A for critical answers: it’s the most deterministic format.
- Keep websites curated: choose stable, canonical URLs and review them periodically.
- Tag consistently: a small controlled vocabulary beats free-form tagging.
A lightweight tagging system
Start with 5–10 tags and resist adding more unless you can explain the new category:
policypricingsupportbillinglegalonboarding
Lifecycle: active vs archived
Knowledge items have a status:
active: eligible for selection and use in flows/botsarchived: kept for history but should not power new automation
Archiving is preferred over deletion when you want an audit trail of what changed.
Review cadence
- Monthly: scan for outdated pricing/hours/refund rules.
- After major product changes: update Q&A first, then websites/files.
- After workflow changes: verify the sources referenced by Knowledge bot nodes still match intent.
Next: