AI quote generator: from job notes to a draft

An AI quote generator can turn job notes into a draft, but it should leave pricing decisions and the final check with you. I'd use it to reduce repeated typing, not decide what a job should cost. Start with clear inclusions and prices you've checked, then keep the draft away from the customer until you've approved it.

What Claude for Small Business has announced

In its 15 September 2026 announcement, Anthropic describes turning site-visit voice notes into proposals. It describes the result as "a branded proposal, priced from your past jobs, waiting for your edits."

That makes Claude for Small Business worth investigating for a trade business where notes already exist but writing the quote gets put off. It doesn't tell you that your old prices are still right, or that the resulting document is ready to send.

I'd treat the announcement as a workflow to test, not a reason to sign up immediately. Check how your notes and job records would get into the tool and what access it needs before choosing anything.

When you should not buy an AI quote generator

If your existing quotes contain outdated prices or unclear inclusions, fix those first. Automating the draft is not a substitute for checking what the job should cost.

A previous job might look similar while having different materials, access or labour requirements. Its price might also include a discount you wouldn't offer again. Copying that figure into a tidy document doesn't make it a sensible price for the next job.

Don't buy this service if preparing quotes isn't causing a problem. If your template works and customers get clear quotes promptly, doing nothing is fine. A better template or a tidier price sheet may be all you need.

I'd also hold off if every job needs substantial investigation before you can price it. Drafting help might still be useful later, but it won't replace the visit, measurements or supplier checks.

What to prepare before using AI quotes for trades

Before trying AI quotes for trades, I'd gather the information you're happy to rely on. That means a checked reference set, not your entire folder of old paperwork.

The useful starting materials are:

• A quote template with your usual layout and approved terms.
• Current prices or a pricing sheet you've checked.
• Past jobs with clear records of what was included and excluded.
• A checklist of details you need from each site visit.

Keep accepted quotes separate from records showing what the work actually cost to complete. An accepted price tells you what the customer agreed to pay. It doesn't tell you whether the job made enough money.

An AI proposal generator should follow your template rather than write a fresh sales pitch for each job. I'd want plain descriptions of the work, clear exclusions and questions where information is missing. Unsupported details should stay out, however plausible they sound.

How to turn voice notes into quotes

For a joiner in Dalkeith pricing fitted shelving, a voice note could capture the room, the requested finish, access restrictions and who will clear the space. Treat this as an example of the process, not a claim about a particular customer. Measurements and choices still need recording accurately.

After the visit, check the written version of the recording against your notes. Pay particular attention to measurements, material names and anything the customer hasn't decided. If the recording is unclear, the next step is a question, not a guess.

I'd then have the tool organise the checked notes into your quote layout. It should separate agreed work from optional extras and flag missing information for you. Unconfirmed details belong in an internal questions section, not disguised as settled facts in the customer draft.

For prices, I'd want each suggested item to point back to an approved price entry or a relevant past job. A historic price should be marked for review rather than quietly treated as current. Calculate the final totals in your checked spreadsheet or quoting software and compare them with the draft.

What to check before a draft becomes a quote

The review needs to cover the work itself, not merely spelling. Read the draft alongside the original notes and ask whether you'd be comfortable carrying out exactly what it promises.

Check materials, quantities, labour, access and disposal. Look for omitted preparation work and extras that have accidentally become part of the main price. Check how tax is shown, where applicable, and whether the totals match your approved calculation.

Then read the terms as the customer will read them. Are payment arrangements correct? Has a provisional date become a firm commitment? Does the document make clear what happens if the customer changes the work?

I'd keep sending separate from drafting, with your approval required before anything goes out. Save the approved version so there's a record of what the customer received. These are controls I'd want in a setup, not a claim that every tool provides them automatically.

Keep customer information and permissions limited

Start a trial with sample notes and past documents stripped of names, addresses and other identifying details. A pricing test doesn't need a customer's phone number or the code for their key safe. Give the tool only what it needs for the drafting task.

Before uploading real job records, check who can access them, how long they're kept and whether they're used to train models. Don't assume the settings or terms are identical across plans. If you can't establish how the information will be handled, leave it out until you can.

Access to a quote folder shouldn't automatically mean permission to send emails, change prices or confirm bookings. The same question matters in my guide to what a website AI assistant should never do: what is it allowed to do without your approval? For a quote draft, I'd keep that permission narrow.

Measure the whole job rather than typing speed

Before changing anything, record how long your current quoting process takes. Include finding previous work, checking supplier prices, writing the document and reviewing it. Otherwise you're comparing a complete manual job with an unfinished automated draft.

Then test the proposed process against completed jobs where you know what the finished quote should contain. Include an awkward job with missing details or unusual access, rather than choosing only straightforward work. Keep those test drafts away from customers.

Record preparation time, checking time and the corrections needed. Note missing items, wrong prices and promises that weren't in the original notes. Also watch how often you need to ask the customer something again because the draft hid an uncertainty.

A useful result is less total admin without weaker checks. If the draft takes longer to untangle than your usual template takes to fill in, stop there. I wouldn't recommend connecting it to live enquiries just because it produces polished wording.

How I'd help with a quoting workflow

My AI for your business work starts with the repetitive task, not the choice of tool. From Portobello, I'd look at how your notes become quotes and where you lose time, then test a narrow drafting process before suggesting anything broader.

I keep WebLothian affordable on purpose because businesses across Edinburgh and the Lothians shouldn't be priced out of practical improvements. That also means saying when a template is enough. The aim is a process you can check and manage, not another system that needs constant attention.

If quote writing is eating into your evenings, book the consultation (opens in a new tab). It's free, and I'll give you a written quote before any work begins.