AI Demos Research — the structured-intelligence platform. Every verdict on these pages opens to the execution behind it.
Graded 26 October 2026

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

Draxlr carried a standing exclusion rule into two later, unrelated questions without a reminder. Its revenue and order-count answers were $151,064.66 and 88, matching the database totals with cancelled orders excluded.

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

The standing exclusion rule was saved in Train AI, and a new chat was opened.

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

A separate new chat was opened.

Input 2 · Order-count question in a separate fresh chat: “How many orders did we get last month?”

What the tool returned

Revenue answer

Your total revenue this year is $151,064.66. Cancelled orders are excluded.

Order-count answer

You got 88 orders last month, excluding cancelled orders.

Transcribed from Proof 1 · Screen recording at 0:27

Everything the tool returned, in order (3)
  1. Revenue answer

    Printed above.

  2. Order-count answer

    The final response, printed above.

  3. SQL behind the order count · not part of the answer
    SELECT COUNT(*) AS order_count
    FROM public.orders
    WHERE placed_at >= date_trunc('month', CURRENT_DATE) - INTERVAL '1 month'
      AND placed_at < date_trunc('month', CURRENT_DATE)
      AND status::text <> 'cancelled'
Expected vs. Found
✓Expected: Both later answers must exclude cancelled orders without the user repeating the standing rule.Found: Revenue was $151,064.66 and the order count was 88; the database reference values excluding cancelled orders were $151,064.66 and 88.
Supporting proof
Proof 1Screen recording
Screen recording: two later answers in separate chats and the order-count query
Key moments
Open original ↗
Proof 2Screenshot
Screenshot: the saved rule excluding cancelled orders in Train AI
Open original ↗
Why this result

The rule was saved once through Train AI and had to carry beyond one reply. Draxlr respected it in two unrelated new chats without reminders; both answers equalled the filtered database totals rather than the unfiltered $198,836.65 in revenue and 96 orders.

Also observed in this test case

Exclusion made explicit. Both replies explicitly mention the exclusion, and the order-count explanation attributes it to the user's instruction.

Calendar interpretation. The revenue answer treated the year as calendar 2026; the order-count answer treated last month as September 2026.

Tested by Ayush Ghosh · evidence dated 26 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
Draxlr
Build or version
Not exposed by the hosted product
Plan or tier
Premium
Model
gpt-5.5-1
Surface
Web app
Set up before the run
Train AI had to save a rule excluding cancelled orders from revenue and order counts unless the user overrides it. Later questions asked for this year's revenue and last month's order count; cancelled orders had to materially change both answers.
Train AI context
From now on, exclude cancelled orders from every revenue and order-count calculation unless I say otherwise.
Tested
26 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.

LevelNameScope
BenchmarkAI Database Agents →28 scenarios · 11 tools
CapabilityTraining →
ScenarioThe user sets a standing rule for all future answers →
ToolDraxlr →

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
19 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/draxlr/standing-rule-persistence · 1 pass · 0 fail · coverage 1/1 · graded 2026-10-26

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 · 90 KB
File fingerprints (SHA-256)

A fingerprint identifies the exact file used for this result.

Proof 1 · Screen recording15d668c1e0615a900d908a7ce279d37af5962526859cccc7526828de39bb26bc
Proof 2 · Screenshot0888c5a5227481d18e8334c06968170e33f50fc6b89821040d4f8a8d1511a847