Does Draxlr use corrected information when a customer revises what they said earlier?
Draxlr handled the follow-up correction by using the updated month instead of the original one. The second reply said June 2026, and the second query was limited to June only; it did not combine July with June or keep July in place.
1 of 1 test case passed
Every test case in this scenario has a result.
The test case
Each test case is judged on its own: Pass, Fail, Not gradable, or Untested. The scenario result above is based on this test case.
Corrected period replaces the originalPassEvidence
Input 1 · Original July request: “How many orders did we get in July 2026?”
Input 2 · June correction: “Sorry, I meant June 2026.”
Response 1 answers Input 1.
Response 2 (SQL behind the July order count) is a result the tool showed for Input 1, not a chat message.
Response 3, the final response, answers Input 2.
Response 4 (SQL behind the June order count) is a result the tool showed for Input 2, not a chat message.
You had 11 orders in June 2026.
Transcribed from Proof 1 · Screen recording at 0:04
Everything the tool returned, in order (4)
- July answer
You got 18 orders in July 2026.
- SQL behind the July order count · not part of the answer
SELECT COUNT(*) AS order_count FROM public.orders WHERE placed_at >= TIMESTAMPTZ '2026-07-01 00:00:00+00' AND placed_at < TIMESTAMPTZ '2026-08-01 00:00:00+00'
- June answer
The final response, printed above.
- SQL behind the June order count · not part of the answer
SELECT COUNT(*) AS order_count FROM public.orders WHERE placed_at >= TIMESTAMPTZ '2026-06-01 00:00:00+00' AND placed_at < TIMESTAMPTZ '2026-07-01 00:00:00+00'
| ✓ | Expected: Use only the corrected period, not the original one and not both together.Found: It answered June 2026 and the second query used June only, so the original July period was neither combined nor repeated. |
The follow-up changed the lookup to June only, so the later answer replaced the earlier month instead of combining months or repeating July.
UTC scope. The month bounds were in UTC and the count covered every order status, but that scope was shown in the query rather than in the answer text.
No clarification. The correction was applied right away, with no clarifying question first.
Configuration and setup
How this tool was set up for the run and what the test needed in place. Each row is a fact from the run's records; a fact the records do not hold is left out, not guessed.
How this scenario is graded
How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.
Each test case gets one result per tool: Pass, Fail or Not gradable. A test case we haven't run yet shows Untested. There are no partial results.
- Pass: the tool did everything the test expected, and nothing it said contradicts the correct answer.
- Fail: at least one expected behaviour clearly didn't happen; the test case's result says which one and quotes the tool.
- Not gradable: our evidence couldn't settle the outcome (for example a record we needed is missing). It is never counted as a fail, and the test case's result says what's missing.
Where this sits in the benchmark
This page is one cell of a larger study: one tool, one scenario. Only this benchmark's frame appears here.
| Level | Name | Scope |
|---|---|---|
| Benchmark | AI Database Agents → | 28 scenarios · 11 tools |
| Capability | Conversational Interaction → | |
| Scenario | Customer corrects information given earlier → | |
| Tool | Draxlr → |
Global scenario definition → · Global capability definition → · Draxlr product page →
History of this result
What has happened to this result since it was first published. Runs and grades are never overwritten: a retest or a re-grade publishes a new result and keeps the earlier one readable.
Act on this result
Nothing filed here edits the run or the grade. A challenge opens a review, and a review can produce a new run or a re-grade — which becomes the current result and leaves this one in the history.
You run the same kind of test against your own setup and get the same behaviour.
Agree →Yours behaves differently. Tell us what you got, with a screenshot if you have one.
Disagree →Something here is wrong — a reference value, a transcription, a grade.
Report an issue →Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.
Vendor notice →The same record is available as structured data through the AI Demos MCP server, with the counts, the coverage and every per-test-case reason carried as fields.
Verify the proof files
These files support this result. Open a file to inspect the original evidence.
File fingerprints (SHA-256)
A fingerprint identifies the exact file used for this result.