Use this article when your tenant is freshly provisioned and you want to run a quote on the template configuration before you start customising machine rates and substrates. After this you will have one fully calculated estimate against a test customer, so you can see how the AI intake and the routing engine work together end-to-end.
Before you begin
A test customer record exists with estimation enabled. See Set up a test customer for onboarding.
You have a sample customer request — paste-able text, an email screenshot, or a PDF RFQ. Required role: MIS User or higher.
Steps
1. Start a new estimate
Open Manage estimates from the GelatoConnect sidebar. Select Create new estimate. The Create Estimate workspace opens with a customer selector, a category selector, and the AI assistant input area.
2. Select the customer and category
Pick the test customer in the customer selector. If required contact fields are missing, the workspace shows Fill all customer details to create the estimate — complete those fields before proceeding.
Pick a product category that matches your sample request. For an A4 stitched book request, pick Stitched Book. If you select a category that has no part selected, the workspace shows Select a product to create the estimate — pick the product (the part inside the category) before continuing.
3. Give the AI the customer request
Use one intake method, then wait for the assistant to populate the product fields.
Paste request text
Copy the request from email, chat, or a portal message.
Paste into the AI assistant input area.
Submit with the send control or by pressing Enter.
Paste or attach an image
Paste an image directly into the assistant input area, or attach a screenshot file.
Submit. The assistant may ask one or two follow-up questions for missing fields like quantity or finished size — answer in the same input area.
Attach a PDF RFQ
Use the attachment control to add a PDF. Submit, then review the extracted fields.
4. Review the populated fields
Scan every field the assistant filled in: size, quantity, page count, colour, substrate, finishing. Correct anything the assistant misread — the assistant should never override an explicit value in the customer's request, but free-form RFQs do leave ambiguity. Expand the Show More rows on the form to confirm advanced fields (binding type, cover stock, finished size) are correct.
5. Calculate the quote
Select Calculate. The routing engine runs against your current Estimate Setup configuration and returns a price breakdown.
If Calculate completes, the price breakdown opens with make-ready, run, paper, finishing, and markup lines, and the route header names the assigned print machine and finishing machines.
If Calculate fails with Unable to calculate estimate, open Issue analysis from the failure banner to see the first missing entity. See Troubleshoot cannot calculate estimate for the symptom-by-symptom fix.
6. Iterate on the spec
Change the quantity from 5,000 to 500. Select Calculate again. The routing engine usually switches from offset to digital at short runs — you should see the assigned print machine change and the make-ready cost share drop. This is the easiest way to confirm the configuration responds to the request rather than echoing a fixed answer.
Things to know
The AI assistant cannot quote capabilities your shop has not configured. If the request asks for spot UV but no spot finishing machine exists in Estimate Setup, Calculate fails. Add the machine or remove the finish from the request before retrying.
Single-product quoting (the default Create Estimate flow) accepts paste, image, and PDF. Spreadsheet RFQs with many lines belong in the Create multi-product estimate flow — open that from Manage estimates instead.
Early quotes use template substrates and template machine rates. Treat the totals as a sanity check on the flow, not as a go-live number, until you have customised your shop's machine rates and substrate prices.