Quote receipt and derivation

Every quote carries a complete replay record — the exact book version, structured inputs, line-item derivation, conditions, and review trail that produced the offer. This record exists so the business can explain, years later, exactly what it offered and why. The same request always produces the same result.

​

What a quote pins

A priced quote records four book pins that together identify the exact artifact that evaluated the request: When all four pins are present, the quote is replayable. When a pin is missing, the receipt shows a Not replayable chip — the stored result and derivation remain visible, but the offer cannot be exactly reproduced against its original version.

​

Structured inputs

Each quote stores the normalized inputs that were evaluated — the service selections, factor choices, and context the customer or staff member provided. These inputs are structured, not free-form text. They are what the Book’s evaluator consumed to produce the result. The same inputs evaluated against the same book version always produce the same result. This is true across HQ, hosted quote paths, embeds, and the public API — they all use the same immutable published artifact and evaluator.

​

Line-item derivation

The derivation table on the quote receipt shows every line item the Book evaluated, in order: Each line item carries its own provenance — the rule that produced it, the factor selections that triggered it, and the amount it contributed. The sum of selected line items is the offered amount.

​

Commercial policy

The policy frozen with the price when the quote was produced: The policy is frozen at evaluation time. If you change the book’s deposit or validity settings later, existing offers keep the policy they were produced with.

​

Review and override record

When a quote is held for owner review, the review record captures the complete decision trail: This trail is the correction history. It shows what the Book produced, what the owner changed, and why. An adjustment classified as an existing rule problem or a missing rule also drafts a learning proposal against the pinned book version — the reason becomes the owner evidence for that proposal.

​

Why the result can be replayed

A quote is replayable when:

1. The book version ID and artifact hash are present.
2. The structured inputs are stored.
3. The evaluator is deterministic.

Given the same book version and the same inputs, the evaluator always produces the same result. This is a structural guarantee, not a promise — the evaluator does not use randomness, external lookups, or model inference at evaluation time. The LLM never sets the number.

​

What this means for owners and developers

For owners, the receipt is the audit trail. When a customer asks why a price was what it was, the receipt shows the exact rules, line items, and conditions. When a correction is needed, the adjustment and its reason are recorded permanently. For developers, the receipt is the replay contract. The same request submitted through the API produces the same result as the same request submitted through HQ or a hosted path. The book version pin means a quote can be replayed against its original version even after the Live book has moved to a new version.

​

Related