Show what supports the answer

A customer asks for Citrus powder in a 300 g tub. Another asks for something ready to drink. A third wants third-party-tested powder. The Qualifier puts each question beside the same Daily Fuel record and shows what supports the comparison.

These are three authored prompts with explicit criteria. No AI provider receives them, and no model recommends the product. The useful result is a distinction between a matching fact, a conflicting fact and evidence that is missing.

Start with the criteria

The Ask view makes the requested attributes visible before the result. The Citrus question asks about format, flavor and pack size. The ready-to-drink question asks about format. The testing question asks about format and evidence of third-party testing.

DAILY / FORM Daily Fuel is declared as Powder, 300 g per tub, with Citrus and Neutral flavors. They are authored for this experience. The photograph illustrates the product; it is not a label scan or proof that a supplier verified the record. The concept adds no price or serving count.

Read the source beside the result

Review the requested attribute, the declared fact and the reason for the comparison. All three criteria in the Citrus powder question are supported by this local record. That result says nothing about health effects, quality or whether the product suits the person asking.

The word Eligible in the chapter navigation has a narrow meaning: satisfies the criteria displayed here. It is not a platform status, ranking score or prediction that an assistant will name the product.

A mismatch is different from a gap

The ready-to-drink request conflicts with the declared Powder format. The record contains evidence for a different format, so the page can explain the mismatch. It does not substitute another product or declare that the brand has no ready-to-drink range.

The third-party-testing request produces a different result. Powder is supported, but the record supplies no testing evidence. Unknown remains unknown. It is neither proof that the product is untested nor permission to write a testing or certification claim.

That difference matters when reviewing a catalog. A known format can answer a format question. A missing test report requires evidence, not a more confident sentence.

Review without rewriting the record

Choose Compare My Criteria to review the selected question. From Ask, it opens Eligible. From Eligible or Gap, it keeps the current chapter. Inspect the Declared Record opens the source facts.

Changing chapters preserves the question and its review status. A different prompt needs a new review; choosing the same one changes nothing. Navigation alone does not complete a review. Start Again restores Powder and flavor in Ask, clears progress and closes the source details.

No AI or supplier service is contacted, and no real catalog is changed. When enabled, site analytics records completion of a local criteria review once per project visit. That event is not an approval or a recommendation result. There is no purchase path, testing verification or certificate generation.

What this demonstrates

The Qualifier demonstrates a way to make product answers inspectable. Every supported statement has a declared fact behind it. Conflicting facts and absent evidence remain separate, so the interface does not conceal uncertainty inside an apparently complete answer.

The supporting articles explain the research motivation and the different jobs of format records and customer-owned training logs. Observed prompt differences do not establish that editing a field causes recommendation gains.

A production version would need current product records, accountable evidence sources and clear rules for stale or missing information. Start with questions your catalog can answer and make the remaining evidence requests explicit.