Turning a website enquiry into a quote draft can reduce repeated copying, but the draft must distinguish supplied facts from assumptions. A missing measurement or unclear service request should become a question, not a made-up line item.
Begin with one type of quote whose inputs and review process are reasonably consistent.
Define the information required
List the customer details, location, requested work, quantities and timing needed for that quote type. Separate required information from details that can be confirmed later.
Identify the approved source for products, rates and inclusions. Do not ask a language model to remember the business’s current price list.
For a hypothetical maintenance company, the workflow might prepare a draft from an approved service package and a customer’s site details. Unusual access requirements would remain an exception for staff.
Make gaps visible
Show the reviewer which fields came from the enquiry, which came from approved business records and which remain unresolved.
Keep exclusions and assumptions explicit. Do not turn “customer asked about availability” into a confirmed start date.
Our business automation examples explain the wider opportunity. A quote workflow deserves its own checks because the output can create expectations about price and scope.
Route the draft for approval
Give the reviewer the original enquiry, pricing source and draft together. Ask them to check quantities, scope, exceptions, terms and any calculations.
Sample reviewer record
| Line | Value | Source | Assumption | Reviewer decision | Status |
|---|---|---|---|---|---|
| Service package | Gutter clean, single storey | Approved package list v4 | None | Approved | Ready |
| Quantity | Awaiting confirmation | Enquiry did not say | Not assumed | Ask the customer | Waiting |
| Start date | Not set | Customer asked about availability | Not a confirmed date | Leave blank | Waiting |
| Price | From the rate table, version 2026-10 | Price list | Rate current | Check total | Draft |
Use deterministic calculation logic for arithmetic and validate the result. Generating convincing prose is not evidence that a total is correct.
The system should make it clear whether a document is a draft, approved or sent. Keep the final sending step under the agreed business authority.
Test changes and failures
Test an incomplete enquiry, a duplicate submission, an expired price, a changed quantity and a failed document creation. Confirm that none silently becomes an approved customer quote.
Record the price-list version used so an old draft can be recognised when rates change. Assign someone to review drafts left waiting after a failure.
Start with a controlled sample and compare the full handling time, including corrections, with the manual process. Expand only when the team can explain the results and recover from errors.
DonutLabs’ quote automation service can help map these inputs and approvals. Bring representative quotes and the exceptions that experienced staff currently catch.
For the wider planning context, read How to choose your first business AI workflow.
Once the draft route works, the next problem is repeated submissions: see how to prevent duplicate leads in an Airtable and Make workflow.
