Use this article when you have run a first estimate but want to understand what happens between paste and price. After this you will know which phase the AI controls, which phase your configuration controls, and where each phase shows up in the quote workspace.
What quote creation is
Quote creation is the workflow that turns a customer request into a priced estimate. It has two distinct phases. The first phase is intake: the AI assistant reads pasted text, images, or attached PDFs and maps them onto the product fields of the category you chose. The second phase is calculation: when you select Calculate, the routing engine consumes the configured product fields, picks an eligible production route across your machines and substrates, and returns a price breakdown.
The split is deliberate. Intake is conversational and forgiving — the AI will keep working with partial information and ask for clarifying fields. Calculation is deterministic — it consumes the fields you confirmed and produces numbers that trace line-by-line to Estimate Setup.
A worked example
Northgate Press receives an email: "Need a quote for an A4 stitched book, 5,000 copies, Silk 100gsm cover."
The estimator opens Manage estimates → Create new estimate, picks the test customer, and selects the Stitched Book category. They paste the email into the AI assistant.
In the intake phase, the assistant fills the product fields: size A4, quantity 5,000, cover substrate Silk 100gsm SRA1, binding saddle stitch. It does not fill the page count — the email did not say — so it asks. The estimator answers 48 pages. The product form now has every required field.
The estimator selects Calculate. The routing engine matches the spec against Northgate Press' configured shop:
Press: XL105 (5,000 sheets is in the offset sweet spot; the substrate's Offset OK tag allows the route).
Binder: Stitcher 1 (configured for saddle-stitched books with the substrate weight range).
Substrate: the Silk 100gsm SRA1 row in the substrates table.
The price breakdown opens. The make-ready line uses XL105's configured setup time and machine rate (190 EUR/hour). The paper line uses the substrate's price per ton (1,400 EUR/ton for the 100gsm row) multiplied by the calculated sheet count. The binding line uses Stitcher 1's rate. The markup line uses the category's markup model. Sum gives the final price.
Every number traces back to a value Northgate Press' admin entered in Estimate Setup — the AI did not contribute a single euro to the total.
What this affects
The AI intake fields — your text/image/PDF input drives which fields the assistant populates. Changing the customer's request changes the spec.
The routing engine output — the spec from intake plus your current Estimate Setup configuration drives which route the engine picks and how each price-breakdown line is costed.
Iteration speed — once the assistant has populated the spec, changing one field (quantity, substrate, finish) and re-selecting Calculate is fast. You can A/B routes by re-calculating without re-pasting the request.
What this does not affect
The price totals — the AI assistant interprets the request but does not invent or override numbers in the price breakdown. If you suspect a wrong price, the cause is in your configuration, not in the intake.
Already-sent quotes — quote creation runs on demand. A quote you already sent does not recalculate when Estimate Setup changes. Open it and select Calculate again to refresh.
Capabilities outside your setup — the AI cannot quote a finishing step that has no machine configured. Calculate returns
Unable to calculate estimatewith the missing entity named in Issue analysis.