Use case

Get a quote in front of the customer while they still care

AI quote and estimate generation is the job of turning an enquiry, a site visit, or a stack of photos into an accurate number the customer sees while they're still comparing options. In most businesses the person who can price the work — the estimator, the ops lead, the owner — is also the busiest person in the building, so quotes queue behind site visits, supplier calls, and the jobs already won. A workflow automation build doesn't decide the price; it does everything around the pricing decision so the one person who can price the work is only doing the part only they can do.

· Reviewed by Artur Horimoto, Founder & CEO

The job: get a number in front of the customer while they still care

Ask any estimator why a quote took three days and the answer is rarely that pricing is hard. It's that pricing was next on a list behind a site visit, a supplier call, and two jobs already under contract. The person who can put a number on a job — accurately, at a price the business can actually deliver — is nearly always the same person doing the work that pays the bills today, which means quoting gets whatever time is left over once everything more urgent is handled.

That queue has a cost that has nothing to do with whether the price itself is right. A customer comparing several home services providers for the same job doesn't wait for the most accurate quote — they book with whoever put a credible number in front of them first and looked organized doing it. By the time your estimator gets to the enquiry, the decision may already be made, not because your price was wrong, but because it arrived after someone else's.

The manual way vs. the automated way

The manual version of quoting looks similar across most trades and service businesses. An enquiry arrives by phone, form, or a chat with a site visit report attached, and it sits in an inbox or on a clipboard until the one person who can price it finds a free hour. That person then has to go find the site notes, check the current price book, recall or search out what a similar job actually cost last time, and confirm supplier pricing hasn't moved — before writing a single figure down. None of that is pricing judgment. It's paperwork standing between the judgment and the customer.

The automated version does the paperwork first. The system reads the enquiry the moment it arrives, pulls together everything that came with it, and assembles a draft the estimator opens already built — numbers in place, notes attached, nothing left to go hunting for. The estimator's job shrinks to the part that actually needs a person: does this price look right, and does anything here not fit the pattern.

Manual Automated
Where the time goes Finding site notes, checking the price book, chasing supplier costs Reading, adjusting, and approving a draft that's already assembled
Speed to the customer Whenever the estimator finds a free hour A draft ready as soon as the enquiry lands
Who sets the price The estimator, working from memory and scattered sources The estimator, working from one draft built on all of it at once
Unanswered quotes Forgotten once sent Followed up automatically until answered
Unusual jobs Estimated the same way as everything else Flagged and routed to a person untouched

The point of automating this step isn't to take the estimator's judgment out of pricing. It's to stop that judgment from queuing behind data entry.

How a build actually works

Assembling everything a quote needs

A quote is rarely built from one input. It's the enquiry itself, whatever notes or photos came out of a site visit, the current price book, a look at what a similar job actually cost the last time it was built, and a check that supplier pricing hasn't shifted since the price book was last updated. Doing that by hand means opening several different places before writing a single line, and often re-finding at least one of them because the file lived somewhere else.

The system pulls all of it into one place the moment there's enough information to start: the enquiry text, any attached photos, site notes logged separately, the relevant price book entries, and the closest matching past jobs for a similar spec. None of that requires a person to go looking — it's already gathered by the time anyone opens the draft.

A draft for review, not a quote that ships itself

Every draft the system produces goes to a person before a customer ever sees it. That's a deliberate design choice, not a limitation waiting to be lifted. The system is good at gathering material and proposing a starting number from the price book and comparable past jobs; it has no way of knowing that the customer mentioned an access problem on the phone that never made it onto a form, or that a particular supplier has been running late this month.

Drafting for review means the estimator's time goes almost entirely into judgment — reading a quote that's already built and deciding whether it holds up — rather than into assembling one from scratch.

The boundary that matters

Human-in-the-loop sounds like a hedge until you think about what a quote actually is. Once a customer accepts it, a quote is a number the business is bound by. Get it wrong in the business's favor and the customer feels overcharged and calls someone else next time. Get it wrong the other way and the business does the job at a loss it can't get back.

That's why the boundary is fixed rather than adjustable: the system prepares, and a person prices. It gathers, drafts, checks the number against the price book and past comparable jobs, and flags anything that looks off — a spec well outside the usual range, a supplier cost that's moved since the price book was last updated, a job with no close match in past work. What it never does is commit a final number to a customer on its own.

Handling variants and options

Very few quotes are one number. Most are a base price plus a set of options — a better material, an add-on, a faster timeline — and the customer needs to see the real difference between them, not a single figure that hides what's included. Building those variants by hand means running the same price-book lookup two or three times with small changes each time, exactly the kind of repetitive step that eats an estimator's afternoon without needing their judgment at all.

The system builds the variants alongside the base quote — same price book, same comparable jobs, same supplier costs, just priced against a different spec — so the estimator reviews a complete set of options at once instead of building each one from a blank page.

Automatic follow-up: where the recoverable money actually is

A quote that goes out and never gets a reply isn't a lost job yet — it's an open one nobody is tracking. Most businesses have no systematic way to chase a quote that's gone quiet, because whoever sent it has already moved on to the next enquiry, and a manual follow-up habit depends on someone remembering to do it days after the fact.

This is usually where the most recoverable revenue in the entire process sits, because a follow-up costs almost nothing while an unanswered quote sitting in an inbox costs the whole job. The system tracks every quote it helped produce, follows up on a sensible schedule if there's been no answer, and hands the thread back to a person the moment the customer replies with a real question rather than continuing to send automated messages into a live conversation.

Knowing when a job is too unusual to template

Not every job fits the pattern the price book and past jobs describe, and pretending otherwise produces a confidently wrong number. A spec nobody has quoted before, a site with a genuine complication, a combination of options the price book was never built to cover — the system's job in those cases isn't to guess. It's to recognize that the input doesn't match anything it has a reliable basis for pricing, and route the enquiry to a person with everything gathered so far attached, untouched by an automated first pass at a number.

That routing decision matters as much as the drafting itself. A system that quietly forces an unusual job through the standard template produces a wrong price with real confidence behind it, which is worse than no draft at all.

What it connects to

A quoting system is only useful if it reaches into the places a quote's inputs and outputs actually live. Typically that means:

  • Your price book or estimating software, read directly so quoted numbers reflect current pricing rather than a spreadsheet someone forgot to update.
  • Your CRM or job management system, both for pulling up past jobs of a similar spec and for writing the finished quote back as a real record — this pairs naturally with keeping CRM data accurate in the first place.
  • Supplier price lists or supplier portals, so material costs in a quote reflect what things actually cost this week, not whenever the price book was last touched.
  • However enquiries and site notes arrive — a web form, a phone call, a technician's photos from a site visit — read the moment they land rather than manually re-entered later.
  • Your invoicing process, since an accepted quote becomes the basis for the invoice; pairing this with automated invoice processing closes the loop from first enquiry to paid job.

None of this requires replacing tools your team already uses. Where a system exposes a clean way in, the build connects to it directly; where it doesn't, there's usually still a workable route in — an export, a shared inbox, a webhook the platform already supports.

Frequently asked questions

Will the system ever send a quote without a person checking it first?

No. Every draft goes to a person before a customer sees it. The system assembles the quote — pricing, notes, comparable past jobs — but a wrong price is a contract you're bound by, so committing the final number to a customer stays a human decision on every build we make.

How does it handle jobs that don't fit our usual pricing?

It recognizes when a job has no reliable basis in the price book or past comparable work and routes it straight to a person with everything gathered so far attached. It doesn't force an unusual spec through the standard template and guess at a number that looks confident but isn't grounded in anything real.

Does this replace our estimator?

No. It removes the data-gathering and drafting work — finding site notes, checking the price book, pulling comparable jobs — so the estimator's time goes into the part that needs judgment: does this number hold up, and does anything here not fit the pattern.

How long does a build like this take to go live?

Most quoting systems are in production within weeks. We usually start with the drafting step — assembling and pricing a quote for review — and add follow-up, variants, or deeper CRM integration once the first piece is proven against real enquiries rather than a demo.

What does it cost?

It depends on how many systems it needs to read from — price book, CRM, supplier pricing — and how much of the drafting and follow-up work it takes on. Every engagement gets a clear price agreed before any build work starts, so there are no surprises once work begins.

Bring us the quote process that's currently queued behind your busiest person's calendar, and in a free 30-minute strategy call we'll tell you honestly what a build like this would take, and what's worth automating first.

Automate this job

Walk us through how it works today. We will map the build and give you a clear price before anything starts.

Free 30 minutes. No pitch deck. You leave with a plan either way.