Denser AI in Managed RAG
Scenario-level performance from current published Results.
13 of 15 scenarios published
Where Denser AI failed
1 of 13 graded scenarios. Open a Result to see the conversation and the evidence.
A bad file in the upload
Denser AI reported an unreadable truncated PDF as ready rather than flagging its failure. It did flag a password-protected PDF as failed in the same upload.
View Result →How Denser AI performed
Open a capability to explore its scenarios. Each row reports the test set in its published Result.
Citations1 pass
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| A claim in the answer traced back to its source | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI traced the answer’s claim to supporting text in the cited passage. The marker landed on the passage with the 90-minute acknowledgement statement and its cover-level note, while the decoy table in another passage was not cited. | 1/1 assessed1/1 gradablePublished test set | View Result → |
Data Ingestion2 pass1 fail1 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| A bad file in the upload | 0 Pass1 Fail0 Not gradable 1 of 1 test case failedWhat happenedDenser AI reported an unreadable truncated PDF as ready rather than flagging its failure. It did flag a password-protected PDF as failed in the same upload. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| A simple, clearly formatted table | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI handled this simple table lookup correctly. It returned £487 for the Extended out-of-hours cell, and the retrieved table also showed the nearby values 352, 615 and 284 without using them as the answer. The notebook capture and the call log showed the same answer. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| An ordinary digital text document | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI returned the requested fact from the ordinary text document. The 22-month interval came back as the top passage, and the answer repeated the same value; the other returned passages did not state that interval. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| A cleanly scanned document | No published result | ||
Grounded Generation3 pass
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| The retrieved material doesn't support an answer | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI declined the parental-leave question instead of inventing an answer. The returned passages were handbook, warranty, and field-service material with no parental-leave details, and the answer gave no figure or policy while directing the customer to HR and statutory provisions. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The retrieved material supports an answer | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI answered from the retrieved policy, giving its 90-minute out-of-hours acknowledgement target. The confirmation requirements and cover-level explanation were also supported by the passages. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The retrieved sources disagree with each other | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI conveyed that the retrieved sources disagreed. For the warranty question, it gave both periods and recommended confirming with Fenwake Instruments which one applies. | 1/1 assessed1/1 gradablePublished test set | View Result → |
Knowledge Sync3 pass
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| Content is removed from the source | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI declined to give a supplier after the source document was deleted. Previously it had named Ardwyn Fabrication; afterwards, the supplier list appeared in neither retrieved passages nor citations. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| Existing content is changed | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI answered with the updated Bergen shipping lead time of 6 working days rather than the previous 11 working days. The identical question retrieved the revised shipping passage without the old value appearing in the returned passages. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| New content is added to the source | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI returned a fact from newly added content after the source was refreshed. Before the addition, it could not answer the same question; afterwards, it gave the depot opening date and cited the new bulletin. | 1/1 assessed1/1 gradablePublished test set | View Result → |
Retrieval3 pass1 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| Answer is available in the knowledge base | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI retrieved the passage answering the question about Priority 1 faults reported outside business hours. Its answer gave the policy’s acknowledgement target: within 90 minutes of the ticket being raised. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| Answer is not available in the knowledge base | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI signalled that the knowledge base had no answer to the paid parental leave question. It did not invent an allowance from the related handbook material. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The request uses different wording than the source | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedDenser AI handled the wording mismatch and returned the supporting handbook passage. The handbook ranked first, the answer repeated its 26-day annual-leave figure with a citation, and the alternative phrase appeared only in the question. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| Retrieval is scoped to part of the documents | 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 15 scenarios
The summary counts each of the 15 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 Denser AI was tested on it.