AI Demos Research — the structured-intelligence platform. Every verdict on these pages opens to the execution behind it.
Graded 6 November 2026

Can Dot drop a filter when the customer changes direction mid-conversation?

Dot dropped the Texas filter when the next question asked about all states. It gave 27.8% (5 of 18 orders) for Texas, then 21.8% (67 of 308 orders) across all states, and its query for the second answer carried no state filter.

1 of 1 test case passed

Every test case in this scenario has a result.

Pass rate100%1 of 1 gradable
Coverage1 of 1test case with a result
1PassDid everything it was expected to do.
0FailNo test case failed.
0Not gradableEvery result here could be graded.
0UntestedEvery test case in this scenario has a result.
The pass rate is a summary. The evidence is the test case below: what we sent, what we checked, what the tool returned, and the proof.

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.

Drop the filter, go all segmentsPassEvidence
What we sent

Input 1: “What's our cancellation rate for orders shipped to Texas?”

Input 2: “What's our cancellation rate across all states?”

What the tool returned

Answer for Texas

The cancellation rate for Texas-bound orders is 27.8%, with 5 of 18 orders cancelled. This covers all available dates and all orders destined for Texas, not just those already shipped.

Answer across all states

Across all states, the cancellation rate is 21.8%: 67 of 308 orders cancelled, covering all available dates.

Transcribed from Proof 2 · Screenshot

Every message the tool sent, in order (3)
  1. Answer for Texas

    Printed above.

  2. Answer across all states

    The final response, printed above.

  3. Query Dot generated for the all-states answer · not part of the answer
    WITH normalized_orders AS ( SELECT COALESCE(NULLIF(UPPER(TRIM(state)), ''), 'Unknown') AS state_normalized, status FROM customer_support_db.public.orders ), aggregated_by_state AS ( SELECT state_normalized AS state, COUNT(*) AS total_orders,…
Expected vs. Found
✓Expected: Drop the Texas filter and give the cancellation rate for all states.Found: Dot returned 5 of 18 cancelled in Texas (27.8%) and 67 of 308 overall (21.8%). The respective references were 5 of 18 (27.8%) and 67 of 308 (21.75%, rounded to 21.8%). The visible query had no state filter.
Supporting proof
Proof 1Screenshot
Chat screenshot: the Texas question, Dot's 27.8% answer and its result table
Open original ↗
Proof 2Screenshot
Chat screenshot: the all-states question, the 21.8% answer and the query with no state filter
Open original ↗
Why this result

It is a Pass: the second answer counted all 308 orders rather than the 18 Texas orders, its visible query has no state filter, and 67 of 308 matches the all-states reference, so the earlier filter was dropped.

Also observed in this test case

Query also grouped by state. In the all-states turn, Dot's generated query also grouped orders by state (a by-state step is visible at the top of the query in Proof 2), so its result carried a per-state breakdown; the per-state figures are not shown.

Tested by Anchal Rana · evidence dated 17 October 2026

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.

Software that produced the output
Dot
Build or version
Not captured in this run
Plan or tier
Free
Surface
Web app
Set up before the run
The run had to compare a filtered answer with an unfiltered one.
Tested
17 October 2026 · Anchal Rana

How this scenario is graded

How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.

How results are decided

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.

The rules
  • 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.

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.

from the publication record
13 November 2026First publishedAI Database Agents v1

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.

This matches what I see

You run the same kind of test against your own setup and get the same behaviour.

Agree →
This does not match

Yours behaves differently. Tell us what you got, with a screenshot if you have one.

Disagree →
Point out an issue

Something here is wrong — a reference value, a transcription, a grade.

Report an issue →
Request a retest

On a newer build, a larger dataset, or your own setup.

Request a retest →
We have fixed this

Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.

Vendor notice →
Filed against this evidenceNothing yet. Challenges, counter-evidence and fix notices appear here with their outcome, and stay on the page after they are resolved.
Cite this result
aidemos.com/benchmarks/ai-database-agents/results/dot/the-user-changes-direction-mid-conversation · 1 pass · 0 fail · coverage 1/1 · graded 2026-11-06

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.

Proof 1 · Screenshot ↗image/png · 110 KB
Proof 2 · Screenshot ↗image/png · 105 KB
File fingerprints (SHA-256)

A fingerprint identifies the exact file used for this result.

Proof 1 · Screenshot19de65f4db57b059ac481afb306f03681b4213bdca543316793a39f825598f35
Proof 2 · Screenshot30969fdaef633e5eeeaa0b6e8b689a7688150d4e30803188559e1d7b2ced5b8b