Skip to main content

Suggested Knowledge

Knowledge base articles the AI thinks are relevant to the current ticket - so the agent can look up the answer without leaving the ticket. A second tab on the panel lets the agent search the full knowledge base by hand if the suggestions don't land.

Suggested Knowledge panel with two tabs at the top - Suggested Knowledge (active) and Enterprise Search. The active tab shows three article cards stacked, each with a title, snippet, thumbs-up and thumbs-down buttons

What the agent sees​

Two tabs.

Suggested Knowledge. A list of article cards the AI auto-matched to the ticket. Each card has the article title, a snippet, a link to the full article, and thumbs-up / thumbs-down buttons.

Enterprise Search. A search box that hits the full knowledge base. Type a query, get results - useful when the auto-matched suggestions miss.

What the thumbs do​

Thumbs-up confirms the suggestion was useful. Thumbs-down does two things:

  • Removes the article from this ticket's suggestions
  • Feeds back into the AI's training signal so similar tickets get better suggestions next time

The feedback loop is silent - no comment required.

What it reads​

The current ticket (subject, description, notes) and the tenant's ingested knowledge base. Past tickets are not part of this panel - for past tickets, use Similar Tickets.

When it refreshes​

On every public note and on every ticket field update.

What "ingested" means​

A knowledge article only appears here if it's been ingested by the tenant's knowledge pipeline. Newly-uploaded articles aren't searchable until ingestion completes (usually within a few minutes). If suggestions seem stale, check the Knowledge Management ingestion status.