Skip to main content
Knowledge Gaps shows recurring Assistant situations where RAUM lacked enough matching knowledge or confidence. Open Quality & Testing → Knowledge Gaps.

​
What a gap contains

A gap can include:
  • topic;
  • number of affected tickets;
  • example Zendesk ticket IDs;
  • a suggested procedure title;
  • first/last seen dates; and
  • status.
The current statuses are:
  • Open
  • Dismissed
  • Resolved

​
Impact labels

The interface derives an impact label from the number of affected tickets so repeated gaps are easier to prioritize. Use impact as a triage aid, not a substitute for reading the example tickets. One rare ticket can still represent a severe policy failure.

​
Run gap analysis

Assistant organizations can start a gap-analysis pass from the Knowledge Gaps view. The analysis updates the gap list from relevant Assistant ticket history.

​
Create a procedure from a gap

For an open gap, choose Create Procedure. When example ticket IDs exist, RAUM opens the procedure-learning workflow seeded with those tickets. This lets the procedure learner draft from the actual recurring cases rather than from a generic topic label. Review that draft before activation. Ticket history describes what happened; it does not automatically define what should happen.

​
Dismiss, resolve, or reopen

  • Dismiss when the gap is not something RAUM should learn or automate.
  • Resolve when the underlying knowledge/configuration has been fixed.
  • Reopen a dismissed gap when it becomes relevant again.
Do not resolve a gap merely because a procedure was created. Retest the representative tickets first.

​
A practical workflow

1

Open the highest-impact gap

Read the topic and example tickets.
2

Identify the real cause

Determine whether the issue is missing knowledge, bad targeting, a failed connection, an ambiguous procedure, or model/retrieval behavior.
3

Create or repair the source

Use Create Procedure when new policy is needed, or fix the existing procedure/guideline/source when the information already exists.
4

Retest

Run representative tickets in Copilot Mode or Simulation.
5

Resolve the gap

Mark it resolved only after the underlying behavior is corrected.

Procedures

Learn from past tickets and turn repeated gaps into reviewed policy.

AI Coaching

Investigate poorly rated individual tickets in detail.