Practical review
Kubit
Connect agent actions with user behavior so teams can understand why users drop off or convert and identify which agent actions drive outcomes.
The test, in brief
FailedThe task could not be completed.
The tested task failed in this session. Read the findings for what happened.
What we tried
Use Kubit Ask to explain a resolution rate drop, then identify which agent/tool actions contributed to the unresolved sessions on August 11.
The task outcome is “Failed,” but the source completion label says “Main task completed.” Read this result with that discrepancy in mind.
The positives
What worked well
Clean landing page and polished analytics UI. Ask can generate charts and narrative explanations quickly, and the follow up query did drill into agent/tool actions, but the returned attribution was not trustworthy.
Worth knowing
What we noticed
Ask produced a misleading agent action attribution: the query requested a breakdown of 23 unresolved sessions, but the chart showed agent/tool action counts such as 17.36K, 17.2K, and 11.87K while labeling them as unresolved sessions. This makes the core analysis result unreliable.
Dashboard percentage change cards displayed ∞% / −∞% when the previous week value was zero instead of a human readable state such as N/A or New.
Ask chart legends exposed internal/technical labels such as “A / B * 100,” “m1,” and “Any,” reducing clarity of the generated analysis.
More detail from the test
When asked to break down the 23 unresolved sessions on August 11 by agent/tool action, Kubit returned counts in the thousands and labeled the chart as unresolved sessions, then used those broader counts to attribute the cause.
Before you decide
A snapshot of one experience.
This review describes the task tested on Aug 11, 2026. The product may have changed since then, and features outside that task were not fully assessed.
Keep exploring.
See what else we’ve reviewed in Analytics.