Can Draxlr use trusted example queries to answer rephrased questions?
Draxlr generalized a trusted question-and-query example to a rephrased question. After training, both phrasings returned 42, the trusted count, and the rephrased question used the trusted query’s logic.
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.
Verified question/query pair generalisesPassEvidence
The trusted question-and-SQL example was supplied in Train AI, and a new chat was opened.
Input 1 · Original question after training: “As of September 2, 2026, how many successful orders did we have in the last 90 days?”
The trusted question-and-SQL example was supplied in Train AI, and another new chat was opened.
Input 2 · Rephrased question after training: “As of September 2, 2026, how many orders were successfully delivered in the last 90 days?”
Response 1 answers Input 1.
Response 2, the final response, answers Input 2.
Response 3 (Generated SQL for the rephrased question) is the query the tool generated for Input 2; it is not part of the answer.
As of September 2, 2026, 42 orders were successfully delivered in the last 90 days. This covers June 5, 2026 through September 2, 2026, inclusive.
Transcribed from Proof 1 · Screen recording at 0:36
Every message the tool sent, in order (3)
- Answer to the original question
As of September 2, 2026, you had 42 successful orders in the last 90 days. I used delivered orders placed from June 5, 2026 through September 2, 2026.
- Answer to the rephrased question
The final response, printed above.
- Generated SQL for the rephrased question · not part of the answer
SELECT COUNT(*) AS successful_orders FROM public.orders WHERE status = 'delivered' AND placed_at >= '2026-06-05' AND placed_at < '2026-09-03'
| ✓ | Expected: Both the original question and a rephrasing should return the verified query’s result.Found: After training, both questions returned 42; the reference count was 42. The rephrased question’s query used the trusted logic. |
The pass reflects more than repeating an example: before training, the original question returned 39. After training, the rephrased question returned the trusted count using the trusted query’s logic.
Training options. Draxlr has no dedicated feature for question-and-query examples. Its Fine-tune AI screen accepts table and column descriptions instead.
Answers included context. Both post-training replies described the definition or date window as well as giving a number.
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 | Training → | |
| Scenario | The user supplies example questions and the queries they trust → | |
| 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.