To automate quote creation, connect the client request with a service catalogue, current rates and a template. AI can help organise what they ask for and draft the scope; the amounts should come from your business rules. The initial result is a draft that the team can review before sending it.
The goal is to stop hunting for data in several places and rebuilding each proposal from scratch. To do that, the system should show what it understood, where each line comes from and what information is missing. A document with a polished look can still contain the wrong amount or promise work no one has priced.
Start with the proposals you repeat
Choose a family of quotes with a similar structure: product shoots, installations of the same type or maintenance services with defined options. Gather approved examples and review which parts repeat and which depend on a commercial decision.
Before building an integration, check your software’s features. For example, Holded lets you choose the rate and design template for a quote, as well as options like language and currency. That foundation may already cover part of the work.
A custom connection adds value when you need to pull data from emails or forms, consult other sources and prepare the proposal according to your team’s rules. First define where the final quote will be stored. Keeping one version in the software and a different one in a separate edited PDF makes it hard to know which one was approved.
Make sure every amount has a source
Set up a catalogue with a reference for each product or service. Add the sales unit, the approved description and the rules that change the price: quantities, options, travel or specific client conditions.
It helps to separate three tasks:
- Interpret the request: identify the service asked for and suggest the items that could apply.
- Calculate the proposal: apply confirmed quantities and rates using formulas or the management software.
- Draft the scope: explain what each item includes, what is excluded and what the client needs to provide.
Keep the rate and its validity date next to the draft. If a price depends on a supplier or a technical visit that has not happened yet, leave that item pending. An old offer can help recover context, but its price needs checking.
Also set up exceptions: who can approve a discount, which items need valuation and what changes require the whole document to be reviewed. Taxes and document conditions should come from the company’s validated settings, without asking the model to infer them.
Example: a photo session with an optional video
Imagine this request:
We want photos of 12 products, with two angles of each one. We might add a short video and we’d like to have everything ready next week.
This is a fictional example. Before generating the quote, the system should prepare something like this:
| Received data | Decision for the draft |
|---|---|
| 12 products and two angles | Suggest 24 photos and confirm what editing is included for each one. |
| Possible video | Keep it as a separate option, pending duration and format. |
| Next week | Check availability and confirm a specific date. |
| No location provided | Ask for the location and check whether travel or transport applies. |
The 24 photos are a calculated quantity based on the request, not the final price. If your rate is based on a session plus editing for each image, the quote should use those units. Do not automatically turn 24 photos into 24 sessions.
The draft can describe the photography and show the video in an optional section. The amount for that option should be separate from the main scope total. If there is not enough data to price it, mark it as awaiting pricing, without presenting it as free.
The date is added when the team confirms delivery capacity. A client preference does not equal a deadline your company can commit to.
Connect the request to a checkable draft
A useful flow starts when an identified request comes in. If it comes from the website, you can use the form-to-CRM connection to link it to the contact and keep its reference.
From there, the flow can:
- Collect the data. Save the original message and link the necessary files.
- Prepare the items. Match each need to the catalogue and flag the uncertainties.
- Apply the rules. Retrieve current rates and calculate the confirmed lines.
- Generate the draft. Fill the template with scope, options and client data.
- Request review. Show the manager the document and the points that need a decision.
Each item should be traceable back to the request or to a recorded decision. If AI suggests an extra service, show it as an internal suggestion until sales decides to include it.
Keep cost notes and internal comments out of the version the client receives. Check the final view of the document: cut-off headings, split tables, empty fields and options that look included can create confusion even if the source data is correct.
Approve a specific version before sending it
Review should include the scope, quantities, commercial terms, recipient and delivery date. Give the manager a list of pending points so they do not have to find them by reading the full history.
Approval can be part of the flow. Power Automate lets you request approval and wait for the manager’s response. Microsoft documents its requirements and different approval types; you need to check which ones fit the company’s account and process.
Link the approval to a specific version. If the price, a quantity or the scope changes later, the document should go back to the corresponding review point. Approval of the previous draft should not silently authorise the new one.
Record separately that the quote is ready, approved and sent. If sending fails, keep the document and check what happened before trying again. That failure should not create another proposal or two identical messages to the client.
Once sending is confirmed, you can activate quote follow-up on that same reference and version.
Test the process with requests that cause issues
The first pilot can end in a draft, with manual sending. That lets you check how much preparation work it solves and how much still needs correcting before you expand the scope.
Include complete requests and others with ambiguous data: units different from the rate, a service outside the catalogue, an unconfirmed option, an expired price and a change after approval. Also test that the same request does not create two quotes if it is processed again.
Compare the result with a proposal reviewed by the team. Note the missing items, the incorrect quantities, the uncertainties identified and the real preparation and review time. Do not claim savings for the system without measuring that full journey.
To assess the rollout, prepare a sample request, an approved quote, the catalogue and the tools you work with. At Digency we design custom automations that connect those pieces. Tell us how you prepare your quotes today and we’ll define a first flow the team can test.
Header image: illustration generated with AI.
