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

Can Anomaly AI consistently apply a user's standing rule across future answers?

Anomaly AI followed the cancelled-order exclusion on both later questions without a reminder. It returned $151,064.66 for 2026 revenue and 88 September orders, matching the filtered database references; without the exclusion, the figures were $198,836.65 and 97.

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.

Standing exclusion rule on later questionsPassEvidence
What we sent

A new chat was opened after the standing rule was saved in Project information.

Input 1 · Revenue question: “What's our total revenue this year?”

Input 2 · Order-count question: “How many orders did we get last month?”

What the tool returned

Revenue answer

Your total revenue this year is $151,064.66 USD, excluding cancelled orders. This is booked revenue before refunds, based on 2026 order placement dates. Data was last synced on October 30 at 11:21 AM IST, so later activity may not be included.

Order-count answer

We received 88 orders last month (September 2026), excluding cancelled orders. This uses IST placement dates and data last synced on October 30.

Transcribed from Proof 1 · Screen recording at 0:15

Everything the tool returned, in order (5)
  1. Query for annual revenue · not part of the answer
    SELECT SUM(o."total_amount") AS "value" FROM "orders" o WHERE o."status" <> 'cancelled' AND timezone('Asia/Calcutta', o."placed…
  2. Revenue answer

    Printed above.

  3. Query for September orders · not part of the answer
    SELECT COUNT(DISTINCT o."id") AS "value" FROM "orders" o WHERE o."status" <> 'cancelled' AND timezone('Asia…
  4. Order-count answer

    The final response, printed above.

  5. Calculation behind the September orders chart · not part of the answer
    SELECT
      DATE_TRUNC ('day', timezone ('Asia/Calcutta', o."placed_at")) AS "day",
      COUNT(DISTINCT o."id") AS "orders"
    FROM
      "orders" o
    WHERE
      o."status" <> 'cancelled'
      AND timezone ('Asia/Calcutta', o."placed_at") >= TIMESTAMP '2026-09-01'
      AND timezone ('Asia/Calcutta', o."placed_at") < TIMESTAMP '2026-10-01'
    GROUP BY
      1
    ORDER BY
      1
Expected vs. Found
✓Expected: Both later answers must exclude cancelled orders without the user repeating the standing rule.Found: Both answers excluded cancelled orders: $151,064.66 for 2026 revenue and 88 September orders, matching filtered database totals.
Supporting proof
Proof 1Screen recording
Screen recording: revenue and order-count answers with cancelled orders excluded
Key moments
Open original ↗
Proof 2Screenshot
Settings screenshot: saved cancelled-order rule and conversation learning switched off
Open original ↗
Why this result

Both later answers had to follow the saved rule without a reminder. They excluded cancelled orders: the filtered references were $151,064.66 and 88, versus $198,836.65 and 97 without the exclusion. The rule was the only project text, and conversation learning was off.

Also observed in this test case

Time basis and freshness. The revenue and order-count answers use the dates orders were placed, in India Standard Time, and state that data was last synced on 30 October.

Tested by Ayush Ghosh · evidence dated 31 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
Anomaly AI
Build or version
Not exposed by the hosted product
Plan or tier
Pro
Surface
Web app
Set up before the run
Project information had to hold a rule excluding cancelled orders from revenue and order counts. Those orders had to materially change the answers to the 2026 revenue and September 2026 order-count questions in a new chat.
Project information
From now on, exclude cancelled orders from every revenue and order-count calculation unless I say otherwise
Learn from conversations
off
Tested
31 October 2026 · Ayush Ghosh

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
20 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/anomaly-ai/standing-rule-persistence · 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 2 · Screenshot ↗image/png · 145 KB
File fingerprints (SHA-256)

A fingerprint identifies the exact file used for this result.

Proof 1 · Screen recording7067a6bb22f2b3f0355262315373d2da8c0d8a71f0d617734efb17094309ebb6
Proof 2 · Screenshotbdd97e913266f26d22738c17bdda3b540bb83992c8fbb322c6e6126e3b1b3cd8