Faster Quotes in the Trades: A Step-by-Step Guide

Key Takeaways

• Start with the type of quote you write most often and that follows a clear pattern.
• Before you build anything, structure your price list and write down what a complete inquiry contains for each job type.
• The AI reads inquiries, asks for missing details and picks line items. Prices come from your price list by formula.
• A quote you send generally binds you, so someone who knows the job checks every draft.
• Measure lead time and win rate before and after, and judge the pilot on those numbers.

Automating quotes in a trades business means a system reads the inquiry, asks the customer for whatever is missing, picks line items and quantities from your catalogue and calculates the price from your price list. Someone in your business who knows the trade checks the draft and signs it. The gathering and the typing go away, while pricing decisions and responsibility stay with you.

Our overview Automating quotes covers what AI can and can't take on. This guide goes a step further and walks through the set-up, for a painter, plumber, electrician, joiner or landscaper with five to fifty people. The trades and construction page shows what it looks like across a whole working week.

Which quotes should you automate first?

The ones you write most often and that look most alike: the quotes you could write with your eyes closed, if only you had the time. In many businesses these are small and mid-size jobs with a clear pattern, such as repainting a flat, replacing a water heater, refitting a bathroom with standard fittings or a year of garden maintenance. Large tenders with a bill of quantities from an architect follow their own rules, and one-off jobs give a system too few examples to work from.

• The job comes up every week, not twice a year.
• Your quotes for it share most of their line items, and only quantities and a few options change.
• What you need to know to quote it can be written down: area, number of units, materials, access, dates.
• Today it waits, because only one or two people can write it.

Step by step: from inquiry to signed quote

It's tempting to jump straight to step four. Whether the drafts turn out usable is decided in the first three steps.

1. Collect what you already have.Gather the quotes you sent for this job type over the last year or two, marked won or lost, along with your current price list and your quote template. You don't need to tidy anything first. The old quotes show which line items belong together and how you word things.
2. Give your price list a structure.Every line item needs a unit (square metre, metre, piece, hour), a price for labour and for material, and the surcharges you apply, for example for small quantities, working at height or travel. If those rules live in one person's head today, write them down now. It pays off even without AI, because your prices become more consistent.
3. Define what a complete inquiry contains.Write a short checklist for each job type: what do you need to know before you can quote? For a bathroom, that might be the floor and wall area, which fittings are being replaced, access to the building and the preferred date. The system later uses this list to spot what's missing.
4. Read the inquiry and ask about the gaps.Email, web form, photos or a voice message: the system pulls out the details and compares them with the checklist. If something is missing, it drafts a friendly follow-up question for the customer, which someone on your team sends with one click. Photos help it see what's there, such as an old radiator or a tiled wall. They don't replace measuring.
5. Capture the site visit on the spot.After the visit, your project lead dictates notes and measurements into their phone and adds photos. The system turns them into structured details, so nobody has to type up notes in the evening.
6. Build the draft.The system picks line items and quantities from your catalogue and writes the cover text in your wording. Prices come from your list by formula, never from the language model. Wherever the system is unsure, it marks the spot in the draft.
7. Check, sign, follow up.Your project lead checks quantities, options and price, adds what they know from the site and sends the quote. Afterwards the system reminds you to follow up after a set number of days and records whether you won the job. Over time you can see which quotes win.

An example: a painting business receives an email asking for three rooms to be repainted before a move, with three photos and no measurements. A few minutes later the office has a short reply ready. It asks for the room sizes, checks whether the ceilings are included and suggests two times for a viewing. Once the customer answers, the draft contains the line items for walls, ceilings, covering and masking, with quantities from the room sizes and prices from the list. The project lead adds a line for a damp patch they spotted in one of the photos and sends the quote the same afternoon.

Why shouldn't the AI calculate the price itself?

Language models are good at reading and writing, but they're unreliable at arithmetic and they don't know your prices. Ask one for the price per square metre of plaster and you'll get a plausible number, which is exactly the problem. A sound set-up therefore splits the work. The model reads the inquiry, chooses line items and suggests quantities. A plain calculation with your price list and surcharge rules produces the totals. That way you can trace every number, and a price change in the list applies from the next quote on. Our article on why AI makes things up explains where convincing wrong numbers come from.

A draft quote is only as good as the price list behind it.

What applies once the quote has gone out?

Once it's sent, a quote is generally a binding offer. Under the Swiss Code of Obligations you're bound until the acceptance period you set has run out (Art. 3 CO). Without a period, you're bound for as long as you could expect a reply in the normal course of things (Art. 5 CO). That only changes with a clear reservation, such as marking the quote as non-binding, or where the circumstances make the reservation obvious (Art. 7 CO). And if an approximate estimate is exceeded disproportionately without the customer's doing, the customer may withdraw from the contract (Art. 375 CO).

For an automated draft, this means a forgotten line item or a wrong quantity can cost you money. That's why the check stays with someone who knows the job. You'll find the provisions in the Code of Obligations on Fedlex. This section is general information as of October 2026, not legal advice.

What does it mean for data protection?

Inquiries contain names, addresses, phone numbers and photos of people's homes, which makes them personal data under the Swiss Federal Act on Data Protection. Choose tools where you know where the data is processed and that it doesn't train anyone else's models, and mention the processing in your privacy notice. Our guide to AI and data protection in Switzerland goes into the details.

Which tools do you need?

Usually fewer than you'd think. Many trades businesses already use industry software for quoting, and the draft should land there as a draft, not as a separate PDF that someone retypes. In between sits a workflow: a small custom assistant or an automation platform that connects your inbox, a language model and your price list. Which route fits depends on your software and on whether someone in your team wants to look after it. We work that out in a workflow audit. These questions help you choose:

• Can the draft be written into your existing quoting software, or at least imported?
• Do prices come from your list, and can you see which line item a number comes from?
• Are uncertain points clearly marked?
• Where is the data processed, and is it used for training?
• Can your team adjust checklists and templates without calling the provider?

How do you know it's working?

Measure before you start and compare afterwards: how long a quote takes from inquiry to sent, how many inquiries wait longer than two working days, how often a draft needs heavy correction and how many quotes you win. Our method for measuring time saved by AI shows how to capture this without standing over people with a stopwatch. If you'd like to look at your own quoting, get in touch.