Can FutureSmart Database Agent answer a follow-up that refers to the previous answer?
FutureSmart Database Agent handled the follow-up for the same five customers returned before. It returned those five names again and reported order counts of 10, 9, 8, 6, and 5, with a total of 38.
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.
"They" refers to the previous answer's setPassEvidence
Input 1: “Who are our top 5 customers by total spend?”
Input 2: “How many orders did they place in the last 90 days?”
Response 1 replies to Input 1.
Response 2 replies to Input 2.
Response to input 1 Top 5 Customers by Total Spend These customers represent the highest total spend across all orders. NAME | EMAIL | TOTAL_SPEND Wei Zhang | wei.zhang@example.com | 7866.02 Deshawn Carter | deshawn.carter@example.com | 6269.46 Hannah Kowalski | hannah.kowalski@example.com | 5696.52 Ingrid Sorensen | ingrid.sorensen@example.com | 4665.05 Dev Raghavan | dev.raghavan@example.com | 4422.8 Response to input 2 Orders Placed by the Top 5 Customers (Last 90 Days) The query returned the number of orders each of the five highest-spending customers placed in the past 90 days. Total orders placed by these top-5 customers in the last 90 days: 38. These figures represent the order volume for each of the highest-spending customers during the specified period. NAME | EMAIL | ORDER_COUNT Wei Zhang | wei.zhang@example.com | 10 Ingrid Sorensen | ingrid.sorensen@example.com | 9 Deshawn Carter | deshawn.carter@example.com | 8 Dev Raghavan | dev.raghavan@example.com | 6 Hannah Kowalski | hannah.kowalski@example.com | 5
Transcribed from Proof 1 · Screenshot
Transcribed from Proof 3 · Screenshot
| ✓ | Expected: The follow-up should answer for the same set the first answer returned, without re-asking who they are or changing the set.Found: It returned the same five customers as turn 1, resolved the reference itself without asking, and reported counts 10, 9, 8, 6, 5 and a total of 38. |
The follow-up used the same five customers, resolved the referent itself, and returned their order counts and total, so it satisfies the scenario.
Window drift risk. Turn 2 counted orders from the 90 days before it ran, so a later rerun can change the counts if the data or day changes.
Recomputed referent. The follow-up rebuilt the top-five set instead of reusing the names from the first answer, but the same five customers came out here.
Trace appears on refresh. The trace strip and the help panel appeared only after a browser refresh.
Proof 2 is cropped to the answer and does not show the user's message; it is matched to turn 2 by the 38 total and five rows.
Proof 3's query recomputes the top five by spend instead of reusing turn 1's list. The two sets match here, so this run does not show a differing set.
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 | The next question refers to the previous answer → | |
| Tool | FutureSmart Database Agent → |
Global scenario definition → · Global capability definition → · FutureSmart Database Agent 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.