A buyer emails an RFQ spreadsheet to your Gmail inbox, then replies: “Make that twelve boxes, same price as June.” The attachment still says ten boxes. Your rep needs a draft based on the latest instruction, with the pack size and current price checked.
Bourne can connect that conversation to your customer records and prepare the quote for review. The workflow needs to follow the whole request through revisions and sending, so the team does not have to reconstruct the thread every time someone replies.
Keep the messages and attachments with one RFQ
Keep the Gmail message ID, thread ID and attachment reference with each extracted line. Save the received time and the exact source version. This lets a reviewer return to the buyer’s wording instead of trusting a detached row of data.
For ongoing intake, Google’s Gmail synchronization process starts with a full synchronization and then uses message history for incremental updates. A stale history ID can require a fresh synchronization. Design recovery so that rereading a message does not create another quote.
A practical record might contain:
| Field | Example value | Purpose |
|---|---|---|
| Request reference | RFQ-218, revision 2 | Keep revisions under the same sales request. |
| Source | Message ID plus attachment and row | Let the reviewer inspect the original. |
| Customer | Confirmed ERP account | Separate the sender’s address from the buying entity. |
| Status | Needs clarification | Prevent unresolved lines from appearing ready to send. |
Store these business fields alongside the Gmail message and thread identifiers.
An email thread can contain several different kinds of change. A buyer can confirm a delivery address while engineering sends a new drawing. Engineering revision control, covered by ASME Y14.35, needs a distinct record from the sequence of Gmail messages.
For each quoted part, retain the drawing and revision the reviewer accepted. If a later message introduces a different revision, reopen the affected technical decision without discarding the confirmed address or commercial terms. This gives the rep a precise review task and prevents an unrelated “thanks” reply from changing which specification the system treats as current.
Apply the buyer’s latest quantity and check the requested price
For the fastener request above, use twelve boxes in the draft and retain the earlier quantity as history. Confirm the number of pieces per box from current product data, then find the June transaction to establish which price the customer means.
The earlier rate in quote history explains what the customer means by “June’s price”; the current agreement determines whether you can offer it again. Keep that decision separate from interpreting the spreadsheet, where tabs, hidden rows or pack descriptions can change the requested quantity.
Review the changes before sending the offer
Give the rep a comparison between the buyer’s request and the proposed quote. Mark changes in quantity, alternatives and exclusions. The rep should not need to reread the whole thread to discover why a line differs. The outgoing quote message should explain those differences to the buyer too, especially an alternative or exclusion.
Keep quote creation and email sending as separate recorded actions. If the email fails after the ERP accepts the quote, retry the send step against the existing quote. Do not create another quotation just because a later action failed.
Set up Gmail access for the agreed workflow
Choose the mailbox and the messages that belong in the queue. A dedicated RFQ address is easier to scope than a salesperson’s entire mailbox. A label can select work within the application, but it is not itself an OAuth security boundary.
Google distinguishes permissions for reading mail, managing drafts and sending. Its gmail.compose scope also permits sending; do not describe it as a draft-only permission. Read the Gmail scope reference with your administrator, then enforce your approval rule in the application as well.
Test replies, forwards and failed sends
Include a revised attachment, a forwarded duplicate, an RFQ with no attachment and a request that spans several replies. Record whether the agent used the latest instruction and whether it preserved the original evidence.
Record review time and corrections for the whole request, including recovered failures. A fast first draft does not represent the time needed to finish a quote that required a retry.
Turn your Gmail inbox into a quoting workflow with Bourne
Choose Bourne when reading the attachment is only the beginning. We can connect the thread to your customer records, prepare quote lines from the latest request and route missing information to the person who can resolve it. The reviewer can see what changed before approving the response.
You can then extend that workflow into follow-up and order review on the same platform. Build your Gmail quoting workflow with our team.
Frequently asked questions
Can Gmail automation prepare quotes without sending them?
Yes. Build a draft-and-review workflow and require approval before sending. Check both the granted API scope and the application’s controls: a scope that permits composing may also permit sending.
Can an AI read PDF and Excel attachments from Gmail?
Yes. The workflow can retrieve the attachments and process their contents. Document extraction is one part of RFQ automation; the workflow still needs to match products, apply current pricing and prepare the ERP transaction.
What happens when a customer changes an RFQ in the same thread?
Link the new message to the existing request, retain both versions and recompute affected lines. Require review if the revision changes an approved quote or an outgoing commitment.
Do Gmail labels prevent an app from reading other messages?
No. Labels can filter the messages your application processes, but the OAuth grant defines API access. Ask your administrator to review the actual scopes and mailbox arrangement.
Further reading
ASME Y14.35: engineering document revisions
Scope of the standard for identifying and recording drawing revisions.
Arda Bulut
Arda Bulut is the co-founder and CTO of Bourne and HockeyStack. He leads engineering at Bourne, building the platform people use to create AI products, agents and automations.