Unified knowledge management for Help Center and Fibi AI Agent

Knowledge is currently split between Help Center articles and Fibi AI Agent Training Data. They live in separate dashboards, use different content models, and cannot be searched, maintained, or automated as one source of truth.

Create one Knowledge dashboard for all reusable support content, similar in concept to Intercom's Knowledge workspace.

• Common content types: public Help Center article, internal article, snippet, uploaded document, and external source.

• Choose where each item is used: Help Center, Fibi, and Copilot. An internal or Fibi-only article must never be publicly visible in the Help Center, but should be available to Fibi when answering.

• Unified management: search, filters, folders, bulk actions, and configurable columns for type, collection, audience, publication status, language, author, last updated date, and where the content is used.

• Consistent editing and lifecycle: one formatting model, drafts and publishing where relevant, source freshness, and clear ownership.

• Complete API and MCP coverage: list, search, create, update, delete, publish, and manage metadata for every supported content type.

The goal is to stop treating public documentation and AI knowledge as separate systems. Teams should be able to create knowledge once, control who or what can use it, and maintain everything from one place.