Skip to content
AI in practice

AI for quoting and estimating: what works for UK SMEs

Easy Insight Team ·

AI helps with quoting when it assembles the draft and a person sets the price. Let it read the enquiry, pick line items and quantities from your own catalogue and past quotes, and fill in standard wording. Keep margins, risk and the final send with someone who knows the job. Without a clean price list, AI only quotes wrong faster.

This guide is for owners and managers of UK SMEs that quote for work: trades, installers, manufacturers, wholesalers and professional services. It describes the shape we would recommend, built from Microsoft's and the regulators' published documentation as of November 2026. It is not legal advice.

Why is quoting a good place to use AI?

Quoting has the traits that make a process worth automating: it repeats, it follows rules that someone could write down, and the slow version costs you money. A quote that goes out three days late is often a job lost to whoever replied first. In our guide to measuring AI return on investment we count "quotes sent the same day" as one of the returns that rarely shows up as hours saved but matters just as much.

It is also a process most firms already part-automate badly. The same customer details and line items get typed into a quote, then again into an order, then again into an invoice. Our guide to business process automation for UK SMEs puts that quote-to-invoice handoff near the top of the list. AI adds something new on the front end: reading a messy enquiry and turning it into a structured draft.

You would not be early. The ONS reports that self-reported AI use among UK businesses with 10 or more employees rose from around 12% in late 2023 to around 35% in June 2026, with improving business operations the most common use (ONS, Artificial intelligence in UK businesses: 2023 to 2026, published 20 July 2026). The same release lists difficulty identifying use cases among the barriers. Quoting is one of the clearer ones.

What can AI actually do in a quote?

Break a quote into its jobs and AI is useful for some of them and risky for others.

Quoting job Can AI do it? Who should own it
Read the enquiry (email, form, notes from a call) and pull out what is being asked for Yes, well AI drafts, person checks
Turn a description into line items and quantities from your catalogue Yes, if the catalogue is clean AI drafts, person checks
Find similar past jobs and what you quoted for them Yes AI
Write standard wording, assumptions and exclusions Yes, from your own templates AI drafts from approved text
Set the price and margin Only as a suggestion A person
Judge risk: access, site conditions, an awkward customer No A person
Send the quote No A person

The table is our assessment, not a vendor claim. The pattern behind it is simple: AI is good at finding and assembling, and poor at knowing what a job will really cost you this month. Material prices move, your capacity changes, and some jobs are worth pricing high because you would rather not win them. None of that is in a model.

What does this look like in software you may already own?

Before buying a new tool, check what your existing systems now include. Microsoft's Business Central is a useful example because its design shows where the line sits.

Its sales line suggestions feature, which Microsoft calls a production-ready preview, creates lines on sales quotes, orders and invoices from a plain-language prompt, and can also find products on earlier documents (Microsoft Learn, Suggest sales lines with Copilot, updated 16 November 2026). An administrator has to switch it on first. Each suggestion carries a confidence rating, and nothing is added to the quote until a user reviews the lines and chooses Insert.

The limits are the instructive part. Microsoft's FAQ says the feature fills only the item, number, unit of measure and quantity; the unit price comes from the pricing logic already in the system (Microsoft Learn, FAQ for sales line suggestions, updated 2 October 2026). It is validated in English only, attachments must be CSV files, and it is not supported on premises. Those details are as of November 2026 and may change.

That is the right split for most SMEs: the AI decides what goes on the quote, and your price list decides what it costs. If a tool you are evaluating lets AI invent prices with no link to your own price data, treat that as a warning sign.

Standalone quoting apps for trades and service firms are multiplying, many promising a quote from a short job description. We have not tested them and will not rank them here. Ask any vendor the questions in the risk section below before you sign.

What data does AI need before it can quote?

This is where most of the work is, and it is work you would benefit from even without AI.

  • One price list or catalogue. Every product or service you quote, with a current price and unit. If prices live in several spreadsheets or in one estimator's head, an AI tool will either guess or copy the inconsistency.
  • Searchable past quotes. Ideally with the outcome: won, lost, or never answered. This is what lets AI say "jobs like this usually include these items".
  • Your standard wording. Assumptions, exclusions, validity period and payment terms, approved once and reused. AI should pick from this text, not write new terms.
  • A rule for who prices what. Which jobs anyone can price straight from the list and which need your most experienced estimator. Write it down; it becomes the tool's escalation rule.

If those four are missing, start there. A small AI proof of concept on a month of real enquiries will show quickly whether your data is ready.

What are the risks of AI-drafted quotes?

A wrong quote can become a binding price. Once a customer accepts a quote, it is usually the basis of the deal. That is why a person should approve every quote before it goes out. In our opinion this is the single rule that matters most.

Consumer price rules. If you quote households, the Digital Markets, Competition and Consumers Act rules on price transparency apply. The CMA finalised its guidance in December 2025: an invitation to purchase must show a total price including mandatory fees and charges, and adding unavoidable charges later is drip pricing (CMA, Price transparency guidance (CMA209); summary from Osborne Clarke, 10 December 2025). Make sure the AI-drafted template always shows one total, VAT included, with any mandatory extras inside it. Whether a particular quote counts as an invitation to purchase is a question for a lawyer, not a model.

Customer data. Enquiries contain names, addresses and sometimes photos of people's homes. Check where the tool sends that data and whether the vendor trains on it. Microsoft, for instance, says its Business Central feature does not use your data or prompts to train models. Our guide to GDPR and AI for UK SMEs covers the checks to run.

Confident nonsense. A language model can add an item that sounds right and does not exist in your catalogue. Use tools that only suggest items from your own list, and look at the confidence ratings.

How do you know it is working?

Measure before and after, on the same kind of jobs. Four numbers are enough:

  1. Time from enquiry to quote sent. The headline benefit.
  2. Edits per quote. How much a person changes the AI draft. If it does not fall over a few weeks, the data underneath needs work.
  3. Win rate. Faster quotes should win more, not fewer.
  4. Margin on won jobs. The check that speed has not quietly cut your prices.

These are our suggested measures. Pick a baseline from the last three months before you switch anything on.

Should you buy a quoting tool or build one?

Buy first if your quoting is standard for your trade and your accounting or ERP system already offers line suggestions. Build when your pricing logic is genuinely your edge: complex configurations, rules that combine materials, labour and travel, or quotes that need data from several systems. Our guide to build vs buy software for SMEs sets out how to decide, and our AI automation service is where we do this work. For the wider picture of where AI fits in a small firm, see AI for business.

Frequently asked questions

Can AI write a full quote on its own?

It can draft one, but it should not send one. AI is good at turning an enquiry into line items, quantities and standard wording. It does not know your margins, your current costs or the risk in a particular job. Let it assemble the draft and keep the price and the send button with a person.

Does AI set the prices on a quote?

In the tools we have checked, usually not. Microsoft's sales line suggestions in Business Central, for example, fill only the item, number, unit of measure and quantity, and the unit price comes from the prices already set up in the system. That is the right design: AI picks what goes on the quote, your price list decides what it costs.

What data do we need before AI can help with quoting?

An up-to-date item or service list with prices, a set of past quotes you can search, and your standard assumptions and exclusions written down. If prices live in one estimator's head or in a dozen versions of a spreadsheet, fix that first. AI will repeat whatever inconsistency it is given.

Do the CMA's price transparency rules apply to quotes?

They apply when you offer prices to consumers. The CMA's guidance, finalised in December 2025, says an invitation to purchase must show a total price that includes mandatory fees and charges. If you quote households, check that AI-drafted quotes show one total including VAT and any mandatory extras, and take advice if you are unsure.

Sources (checked 17 November 2026): ONS, Artificial intelligence in UK businesses: 2023 to 2026 (20 July 2026); Microsoft Learn, Suggest sales lines with Copilot (updated 16 November 2026); Microsoft Learn, FAQ for sales line suggestions with Copilot (updated 2 October 2026); CMA, Price transparency guidance (CMA209), with Osborne Clarke's summary (10 December 2025).


Easy Insight is a UK consultancy for AI, web, apps and data — senior specialists only, no juniors.

Next step

Wondering where AI would actually pay in your business?

An AI strategy and readiness review (from £2,500) tells you where AI pays, where it doesn't, and what to leave alone. The price is fixed in writing before we start, and the advice comes from a team that runs AI in its own products.

Keep reading