Scenario in benchmark · Version 1
The user changes direction mid-conversation
A conversation changes direction, and the agent must answer the new scope rather than carry the old one forward.
How the tools performed
Every in-scope tool is visible. Outcomes come from current published Results for this scenario and benchmark version.
Publication availability: 1 tool has a current published Result.
No published result does not tell you whether a tool has been tested. Test coverage is shown only for comparable published Results.
| Tool | Published outcomes | Test coverage | Result |
|---|---|---|---|
| Published Results · alphabetical, not ranked | |||
| FutureSmart Database Agent | 0 Pass1 Fail0 Not gradable 1 of 1 test case failed | 1 of 1 assessed1 of 1 gradable | View Result → |
| No published result · alphabetical | |||
| AI for Database | — | — | No published result |
| Anomaly AI | — | — | No published result |
| AskYourDatabase | — | — | No published result |
| Basedash | — | — | No published result |
| BlazeSQL | — | — | No published result |
| camelAI | — | — | No published result |
| Definite | — | — | No published result |
| Dot | — | — | No published result |
| Draxlr | — | — | No published result |
| Querio | — | — | No published result |
Assessed = Pass + Fail + Not gradable. Gradable = Pass + Fail. Both use the published Result’s pinned-test denominator. — means not publicly available.
Test design
- Pinned test cases
- 1
- Disclosure
- 0 public · 1 withheld
- Capabilities exercised here
- Conversational Interaction
What this scenario evaluates
- Whether the agent drops the earlier subject or filter when the user changes direction.
- Whether the answer reflects only the new scope introduced by the follow-up.
- Whether the agent avoids silently carrying forward the prior constraint into the new answer.
Exact wording, inputs, fixture state, expected output and detailed grading remain at the test-case level and may be withheld while the benchmark version is active. The scenario and its evaluation intent are public.
How the results are graded
- Pass: the test-case expectations hold.
- Fail: an expectation demonstrably does not hold.
- Not gradable: the evidence cannot establish the outcome.
Version 1 uses test-case expectations; no scenario rubric is pinned.
Benchmark methodology →