How can we help?

The AI performance dashboard

Updated

See exactly how your AI agent is doing — and what to fix next — in one dashboard.

What it shows

  • The engage → resolve → hand-off funnel: how many conversations the AI engaged with, resolved outright, and handed to a person.
  • The grounding rate: how often its answers were backed by a knowledge base article.
  • The top hand-off topics: the questions it keeps passing on — your knowledge base gaps, ranked.

Shown with our demo workspace.

How to read the funnel

A high hand-off rate is not a broken agent. It usually means the questions being asked are not covered by anything you have published, which is a content problem with a content fix.

A low grounding rate is the number to worry about. It means answers are going out with little or no article behind them — exactly the situation where an answer can be confidently wrong.

The most useful screen in the product

The hand-off topics list is ranked by how often each one occurs, which makes prioritizing obvious: the article at the top of that list will deflect more conversations than any other you could write.

From each topic you can draft the missing article in one click and publish it after review. Then watch that topic fall down the list.

CSAT, split by who answered

Satisfaction is shown for AI-resolved conversations separately from human-resolved ones. This is the honest test of whether automation is helping your customers or just your queue — and it is the number to check before turning the agent up.

Everything traces back

Every figure comes from your own conversations. Nothing is estimated, and any number can be followed back to the threads that produced it.

Was this article helpful?