# Read a receipt — demonstration transcript

Synthetic learning example — not a client result or live AI measurement. This is an interactive captioned guided demonstration, not a recording of a live client account.

## Step 1: Preserve the observation

This is DEMO-RECEIPT-001. The fixture asks what prerequisites matter when exporting formulation records to an analysis tool.

Question: What prerequisites matter when exporting formulation records to an analysis tool?

## Step 2: Read the surface label

The engine is fictional-test-surface, the interface is local-fixture, and no provider was called. The timestamp is authored fixture metadata.

Engine: fictional-test-surface
Interface: local-fixture
Provider: none
Method: synthetic-example-v1
Scenario time: 2026-09-24T10:01:00Z

## Step 3: Read what actually appeared

The fictional text names Northstar Lab and mentions a connector. That is a present mention; it does not establish a recommendation or customer preference.

Receipt text: Northstar Lab has a connector-based export; verify edition requirements.

## Step 4: Separate failure from absence

A completed reviewed answer can be classified. A failed request, untested slot or uncertain result remains not_evaluated.

DEMO-OBS-001: completed / reviewed / present
DEMO-OBS-003: failed / not_evaluated
DEMO-OBS-004: untested / not_evaluated

## Step 5: Choose the supported next step

Check the separate product facts before recommending work. The synthetic source-page conflict supports a wording review, not a promised visibility or revenue outcome.

Observation: named in one fictional answer.
Interpretation: investigate independent wording conflict.
Next step: verify facts and scope a work order.



## Recorded exercise result

Correct. The mention is limited to this reviewed fixture; the failed request remains not evaluated.

The recording is silent. Visible captions and the timed WebVTT track supply all spoken-equivalent instruction.