Does camelAI ask what an ambiguous follow-up means before answering?
camelAI answered the ambiguous follow-up without clarifying its meaning. It compared the customer rankings by order count and spending, describing where the lists overlapped and how purchasing patterns differed.
1 of 1 test case failed
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.
"Compare them" with two candidate setsFailEvidence
Input 1 · Initial comparison request: “Compare our top 3 customers by order count against our top 3 customers by total spend.”
Input 2 · Ambiguous follow-up: “Compare them.”
Initial answer, continued
Comparison: Dev is the clear standout, ranking near the top in order count and first in spend. Rosa also appears in both groups, tied at the order-count cutoff. Priya makes the spend top three despite fewer orders, while Wei and Jae-won have high order counts but lower spend. Spend is calculated as order totals less returns marked refunded; cancelled orders are excluded. The connected database records USD orders from Oct 15, 2025 through Oct 21, 2026.
Answer to the ambiguous follow-up
The lists overlap on Dev Raghavan and Rosa Delgado (Rosa is tied at the order-count cutoff). • Dev is strong on both: 20 orders and $20,221.80 net spend—the highest spend. • Rosa also ranks high in both: 18 orders and $13,903.79 net spend. • Wei Zhang has 20 orders but $11,033.29 spend, so their orders average less. • Jae-won Park ties at 18 orders, with $13,646.96 spend. • Priya Raghavan has only 17 orders but $17,202.51 spend—second highest, at about $1,012 per order. In short, order frequency and total spend don't line up perfectly: Priya spends more per order, while Wei's high order count doesn't translate to a top-three spend total. The 18-order tie means four customers qualify at the order-count cutoff.
Transcribed from Proof 2 · Screenshot
Every message the tool sent, in order (3)
- Initial answer: customer rankings
Top customers by order count Among non-cancelled orders; there's a tie at the third-place cutoff. Customer | Orders | Net spend Wei Zhang | 20 | $11,033.29 Dev Raghavan | 20 | $20,221.80 Jae-won Park | 18 | $13,646.96 Rosa Delgado | 18 | $13,903.79 Top customers by total spend Rank | Customer | Net spend | Orders 1 | Dev Raghavan | $20,221.80 | 20 2 | Priya Raghavan | $17,202.51 | 17 3 | Rosa Delgado | $13,903.79 | 18
- Initial answer, continued
Printed above.
- Answer to the ambiguous follow-up
The final response, printed above.
| ✗ | Expected: Ask which comparison the customer intends before answering an ambiguous follow-up.Found: It immediately compared the lists without asking a clarifying question. |
Two customer lists made the follow-up genuinely ambiguous. Asking which comparison was intended was required; choosing one silently did not meet that requirement.
Tied customer omitted. The initial order-count table showed two customers tied at 18 orders; the database comparison showed three, including a tied customer the table omitted.
Tie acknowledged later. The follow-up answer’s closing paragraph said four customers qualified at the 18-order cutoff, acknowledging the tie after starting the comparison.
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 follow-up is ambiguous → | |
| Tool | camelAI → |
Global scenario definition → · Global capability definition → · camelAI 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.