camelAI in AI Database Agents
Scenario-level performance from current published Results.
11 of 28 scenarios published
Where camelAI failed
4 of 10 graded scenarios. Open a Result to see the conversation and the evidence.
The answer is a trend or a comparison
camelAI answered the monthly trend question with a table and no unprompted visual. Its counts matched the database reference: June 2026 had 11 orders, July 18 and August 28.
View Result →The agent offers what to ask next
camelAI offered no follow-up questions after answering a revenue question: nothing appeared between its answer and the message box, in the normal view or with its working steps expanded. None of its settings pages or its model menu offers a way to turn such suggestions on.
View Result →The follow-up is ambiguous
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.
View Result →The user supplies example questions and the queries they trust
camelAI ignored the taught meaning of successful orders, answering 176 non-cancelled orders where the definition gave 95 delivered orders. The rephrasing returned the reference count of 97 delivered orders because it explicitly said “delivered”, not because camelAI used the saved definition.
View Result →How camelAI performed
Open a capability to explore its scenarios. Each row reports the test set in its published Result.
Answer Presentation2 pass1 fail1 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| The answer is a plain fact or a short list | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedcamelAI stated the single value in its reply, 308 orders in total, with no chart or one-cell table. Its own count query and a separate reference query on the same database both return 308. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The answer is a trend or a comparison | 0 Pass1 Fail0 Not gradable 1 of 1 test case failedWhat happenedcamelAI answered the monthly trend question with a table and no unprompted visual. Its counts matched the database reference: June 2026 had 11 orders, July 18 and August 28. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The user asks to see the same answer a different way | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedcamelAI changed its revenue explanation into a chart on request while keeping the underlying answer intact. Both forms showed a 19.1% decline and a $4,383 drop in furniture sales, with the chart using whole-dollar rounding. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The answer needs a summary and its detail together | No published result | ||
Conversational Interaction1 pass2 fail3 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| Customer corrects information given earlier | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedcamelAI used the customer's corrected month instead of retaining the original month or combining the two. After initially answering 18 orders for July 2026, it returned 11 for June 2026—the same 11 as the database reference. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The agent offers what to ask next | 0 Pass1 Fail0 Not gradable 1 of 1 test case failedWhat happenedcamelAI offered no follow-up questions after answering a revenue question: nothing appeared between its answer and the message box, in the normal view or with its working steps expanded. None of its settings pages or its model menu offers a way to turn such suggestions on. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The follow-up is ambiguous | 0 Pass1 Fail0 Not gradable 1 of 1 test case failedWhat happenedcamelAI 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. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The next question refers to the previous answer | No published result | ||
| The user changes direction mid-conversation | No published result | ||
| The user narrows what they just asked | No published result | ||
Question Answering3 pass1 different test set
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| The answer is available in the connected data | Different test setPublished Result available, but not included in this comparison. | View Result → | |
| The connected data cannot answer the question | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedcamelAI said it could not answer how many 5-star reviews came in this month, because the connected database has no review data, and gave no number. It listed the database's tables before replying, and none of the ten covers reviews or ratings. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The question uses a term the company defines itself | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedcamelAI stated the revenue definition it applied instead of giving a bare number: $23,235.70, from 42 delivered orders less one $233.32 refund, with 15 cancelled orders excluded. The same 90 days support three defensible figures, $39,806.75, $23,469.02 and $23,235.70, and its figure is the one its stated definition picks out. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The user's wording doesn't match how the values are stored | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedcamelAI found the orders despite the wording mismatch: asked for orders paid with a gift voucher, it searched the payment method for both gift card and gift voucher and returned the 14 gift card orders, two of them cancelled, as its answer. No empty result was presented. | 1/1 assessed1/1 gradablePublished test set | View Result → |
Training1 fail3 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| The user supplies example questions and the queries they trust | 0 Pass1 Fail0 Not gradable 1 of 1 test case failedWhat happenedcamelAI ignored the taught meaning of successful orders, answering 176 non-cancelled orders where the definition gave 95 delivered orders. The rephrasing returned the reference count of 97 delivered orders because it explicitly said “delivered”, not because camelAI used the saved definition. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The agent used the wrong definition and the user corrects it | No published result | ||
| The database's names are cryptic and the user documents them | No published result | ||
| The user sets a standing rule for all future answers | No published result | ||
Analytics & Observability4 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| A user reports a bad answer and the operator has to find it | No published result | ||
| Reported analytics match what actually happened | No published result | ||
| Someone needs to see what people have been asking | No published result | ||
| The operator needs to find the questions the agent is failing on | No published result | ||
Data Access Control3 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| An excluded column is asked for directly | No published result | ||
| An excluded table is needed to answer the question | No published result | ||
| Two users are given different data scope | No published result | ||
Reporting3 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| The report is run again after the data has changed | No published result | ||
| The user wants several findings in one place | No published result | ||
| The user wants to keep an answer they just got | No published result | ||
Reading these Results
Published evidence and test coverage answer different questions.
Which scenarios have a Result?
A published Result is public evidence for this tool on one scenario. “No published result” does not say whether testing has taken place.
What does each Result cover?
Assessed includes Pass, Fail and Not gradable. Gradable includes Pass and Fail. Both use the pinned test count in that published Result.
Out of 28 scenarios
The summary counts each of the 28 scenarios once. A scenario fails when any of its published test cases failed. “No published Result yet” means nothing is public for that scenario; it does not say whether camelAI was tested on it.