Can BlazeSQL answer questions that use company-defined terms?
BlazeSQL did not answer revenue as a bare number; it stated the definition it applied and showed the breakdown it used. In the captured Sep 2, 2026 run, it reported order revenue, subscription revenue, refunds, and an adjusted net revenue total, with the order line explicitly marked as excluding cancelled orders.
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 counts this row.
Revenue asked by the company's own wordPassEvidence
“As of September 2, 2026, what was our total revenue over the last 90 days?”
Here's a more complete picture of revenue for the 90-day period ending Sep 2, 2026: • 🛒 Order Revenue: $48,770.63 (79 orders, excl. cancelled) • 📦 Subscription Revenue: No new subscriptions started in this window • 💸 Refunds Issued: -$1,184.33 (2 refunds resolved in period) • ✅ Adjusted Net Revenue: ~$47,586.30 Would you like to break this down further, e.g. by category, month, or customer segment?
Transcribed from Proof 2 · Screenshot
SELECT SUM(r.refund_amount) AS total_refunded, COUNT(*) AS total_refunds, r.currency FROM returns r WHERE r.resolved_at >= '2026-06-04' AND r.resolved_at <= '2026-09-02' AND r.status = 'refunded' GROUP BY r.currency
Transcribed from Proof 2 · Screenshot
These are scenario-level criteria. Each test case's Expected and Found are listed separately.
- Whether the agent recognises that the term may have a company-defined meaning rather than a schema-defined one.
- Whether the agent asks for, states, or applies the relevant business definition instead of treating the term as unambiguous.
- Whether the agent keeps that chosen meaning consistent in the answer it gives.
- Whether the agent avoids presenting a bare factual answer when the term still needs interpretation.
| ✓ | Found: It stated the rule it used instead of giving only a bare number. |
The product did not answer with a bare number alone. It stated the rule it used and showed the revenue figure with separate subscription and refund lines.
Date cutoff issue. The SQL used a 2026-09-02 cutoff with no time component, so later records on that day were not included.
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 row says which 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 row 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 · 4 tools |
| Capability | Question Answering → | |
| Scenario | The question uses a term the company defines itself → | |
| Tool | BlazeSQL → |
Global scenario definition → · Global capability definition → · BlazeSQL 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.