Skip to main content
The knowledge base is your team’s article library — SOPs, how-to guides, and troubleshooting steps — organized into categories, tagged, and full-text searchable. Articles can be drafted by AI from a plain-language prompt, and relevant ones surface automatically while you work a helpdesk ticket. Open it at /kb.

​
When to use it

  • Document standard operating procedures, how-tos, and fixes your techs reuse.
  • Publish customer-facing self-service articles (keep others internal-only).
  • Surface relevant articles automatically inside a helpdesk ticket.
  • Draft a first pass with AI, then edit and publish it.
  • See which articles get viewed and marked helpful most.

​
Writing and publishing

Each article has a status — Draft, Published, or Archived — and an auto-generated slug from its title. Organize articles with:
  • Categories — hierarchical (a category can have a parent) with per-category article counts.
  • Tags — comma-separated, for cross-cutting topics.
  • Internal vs. public — mark an article internal to keep it staff-only.
Only Published articles appear in search; drafts and archived articles stay out of results. Search from /kb/search. It uses PostgreSQL full-text search over the article’s indexed content and returns results ranked by relevance with a highlighted snippet showing your terms in context. Multi-word queries match articles containing all the words. Results are scoped to your tenant and limited to the top matches.

​
AI article generation

You can have AI draft an article from a short prompt:
1

Describe the article

Give a plain-language prompt (e.g. “How to reset a user’s MFA in Entra”).
2

AI drafts it

Claude writes as an MSP technical writer and returns a title, markdown content, and suggested tags.
3

Review, edit, publish

The draft opens in the editor — it is not auto-saved. Edit it, then publish when you’re happy.
AI generation requires your tenant’s Claude API key to be configured in settings. Without it, generation returns “not implemented” rather than a draft.

​
In-ticket suggestions

When working a helpdesk ticket, the knowledge base surfaces the most relevant published articles so a fix is one click away.
These suggestions are search-based, not AI: they run the same full-text search against the ticket context and return the top matches. AI is used only for drafting articles, not for suggesting them.

​
Good to know

  • No article version history or scheduled publishing yet — edits overwrite in place.
  • View counts are best-effort (recorded in the background), so treat them as approximate.