Key Takeaways

  • Quotes are slow because of all the gathering: details, line items, prices. The writing is the quick part.
  • AI is good at reading inquiries, finding similar past quotes and suggesting line items from your catalogue.
  • Prices, discounts, what you notice on site and the final signature stay with your people.
  • Before you start, you need your recent quotes, your price list and one person who owns quoting.
  • Start with one type of quote, measure how long it takes today, and judge the pilot against that.

Why quotes take so long

Ask the people who write your quotes where the time goes, and they'll rarely say the writing. It's the gathering. They read the inquiry and look at the photos, hunt for the quote you did for a similar job last spring, copy its line items, check the prices are still right, add whatever is different this time and put it all into the right format. Usually only two or three people know how to do this well, so quoting gets squeezed into their evenings.

That's also why quotes are a good place to start with AI. The inputs repeat, the building blocks are already sitting in your old quotes and price lists, and someone checks the result before it goes out. You aren't asking a system to make decisions for you. You're asking it to do the gathering.

What AI can take over

  • Reading the inquiry (an email, a web form, notes from the site visit) and picking out the job, quantities, dates and special requests.
  • Finding your most similar past quotes, so nobody has to remember which job it looked like.
  • Suggesting line items and quantities from your own catalogue, with current prices from your list.
  • Drafting the cover text in your wording and pointing out what's missing, like a measurement, a material choice or a question about access.
  • Putting everything into your template, ready for you to check.

What stays with your people

  • Pricing decisions: margins, discounts, and what a loyal customer or a tight schedule is worth to you.
  • The judgement you only get from seeing the site, knowing the customer or knowing what tends to go wrong.
  • Unusual jobs the system hasn't seen before. It should hand those over to you instead of guessing.
  • The final check and the signature. If nobody reads the draft, you've added a risk and saved nothing.

What you need before you start

  1. Your recent quotes.A good set of quotes you actually sent, ideally with the accepted ones marked. They show the system your line items, your wording and how you lay things out.
  2. Your price list or catalogue.Whatever you price from today. A spreadsheet is fine. The draft should never make up a price.
  3. Where your inquiries arrive.Email, a web form, phone notes or a portal. The pilot needs to read them where they already land.
  4. One owner.The person who writes most of your quotes today. They decide what a good draft looks like and whether the pilot works.

A realistic first pilot

Pick one type of quote that comes up often and follows a pattern you'd recognise. Before we build anything, measure how long a quote of that type takes you today, from inquiry to sent, and how many inquiries wait longer than a day. That's your baseline, and it's what the pilot has to beat. We describe how to measure the work in our five-step workflow audit.

Then run the pilot on current inquiries for a fixed period, with the owner checking every draft. At the end you'll know how much time a draft saves, how often it needs heavy editing and whether your team trusts it. That's when you decide whether to roll it out, and whether Artifex builds it for you or your team learns to run it. The faster quotes page shows what this looks like in different industries.

Questions to ask any provider

  • Will our prices, customers and past quotes be used to train anyone else's models? Where is our data stored?
  • Does the draft use our templates and our price list, or its own?
  • What happens when the system isn't sure? It should tell you instead of filling the gap.
  • Who checks the drafts during the pilot, and how will we measure whether it saves time?
  • If we stop, what do we get to keep?