AI for home improvement contractors: what to automate first, and what to leave to a person

If you run a remodeling company, build pools, or do any trade where a single job is worth thousands, you've probably been sold "AI" several times this year. AI receptionists, AI estimators, AI marketing, AI follow-up.

Some of it is useful. Some of it will cost you jobs, or get you into legal trouble.

We're an AI engineering company. We build AI systems for clients and run our own product in production, a website assistant called ConGreeto, so read this knowing we sell in this space. We've tried to write the guide we'd want a contractor to read before talking to us or anyone else: what AI does well in a contracting business, what it shouldn't touch, and when to buy a ready-made tool instead of paying for a custom build.

Where contractors are with AI today

The most recent large survey we found is Jobber's 2026 Home Service Trends report: 1,050 US home service business owners, surveyed through Conjointly in December 2025, margin of error ±3 points at 90% confidence. It covered HVAC, plumbing, roofing, cleaning, lawn care, electrical and general contracting.

52% of owners said they use AI in day-to-day operations. Another 27% planned to start within a year, and 37% said they had no plans to.
Among those using it, 54% used it for quoting, 52% for invoicing and 51% for business writing such as emails and proposals.
57% of general contractors used AI, "especially for estimating and customer-facing comms".
On speed: only 11% of HVAC businesses replied to new leads within an hour, against 26% of cleaning businesses.

So contractors who use AI mostly use it for paperwork. That's sensible, but notice where it isn't being used: the first contact with a new customer. For a contractor quoting large jobs, that's where most of the money is lost.

How to decide what to automate

Before any specific tool, use one test. AI is a good fit for a task when the task is:

Repetitive: the same kind of question or document, many times a week.
Mostly words: answering, summarising, drafting, sorting.
Easy to check or undo: a person reviews the result, or a mistake costs little.
Not a promise: it doesn't commit you to a price, a date, a structural judgement or a contract term.

The fourth point matters most in contracting. A wrong answer about your opening hours is a small mistake. A wrong answer about whether a pool fits on a sloped lot, or what a kitchen will cost, can become a dispute.

What to automate first

Ranked by how much they usually save, and how safe they are to hand to AI.

1. Enquiry intake and the first reply

What it does: answers website visitors' questions at any hour, asks the qualifying questions (project type, rough budget band, timeline, home ownership), and saves the visitor's name, number and project notes for your team.

Why it's first: remodeling and pool enquiries are worth a lot each, and many arrive when nobody's in the office. Homeowners often research these projects in the evening and at weekends. An AI assistant that answers and qualifies them at 9 PM turns a lost visitor into a lead your team can call at 8 AM. It also passes the four tests: it's repetitive, mostly words, a person follows up on every lead, and it can be told never to promise a price.

Buy or build: buy. This is a common need, and good products exist. Ours is ConGreeto. Whichever you choose, check that it:

answers from your website and services, not generic text,
saves contact details during the conversation, not only at the end, so an unfinished chat still leaves you a lead,
lets you write rules in plain language ("never quote a final price, give the published range and offer a site visit"),
says plainly that it's an assistant and keeps your phone number visible,
shows you which ad or campaign each lead came from.

One lesson from running this in production. Capturing contact details sounds trivial, but it's where we found real bugs. In our own product, an email address typed at the end of a sentence, as in "you can reach me at jane@example.com.", wasn't being captured, because the final full stop broke the match. That's how most people type an email address. We fixed it and wrote regression tests for it. Whatever tool you use, test it the way your customers actually write before you trust it.

The ConGreeto team covers the remodeling and pool side of this in more depth in remodeling and pool builder leads: why $50,000 enquiries leak before the first call.

2. Drafting quotes and proposals

What it does: turns your site-visit notes, measurements and line items into a clean written proposal: scope, exclusions, payment schedule, timeline.

Why it's useful: it's the most common AI use in the Jobber data (54% of AI users). A well-written proposal that arrives the day after the visit beats a better-priced one that arrives a week later.

The rule: AI drafts, a person prices and signs. The numbers should come from your own price book or estimating software, never from the AI's general knowledge. Check every exclusion and allowance line before it goes out. A model will happily write a confident scope that leaves out who pays for the dumpster.

Buy or build: buy. Most field-service and estimating platforms now include this, and a general AI assistant can draft from a good template. Custom work only makes sense when your pricing logic is unusual and lives in your own data (see below).

3. Follow-up messages

What it does: drafts the one-week and one-month follow-ups on open quotes, the seasonal note to a pool lead who said "next summer", and the request for a review after handover.

Why it's useful: most quotes die from silence, not from a "no". Remembering to follow up is the hard part, and software is good at remembering.

The rule: keep it personal and infrequent. Two useful messages are better than six generic ones. And be careful with texts and calls (see the next section).

4. Review replies

What it does: drafts replies to Google reviews.

The rule: draft only, and personalise every one. In BrightLocal's Local Consumer Review Survey 2026 (1,002 US adults, published February 2026), 89% of consumers expected businesses to respond to reviews, and 50% were put off by templated or generic replies. An AI draft that names the job ("thanks for trusting us with the primary bath, the curbless shower turned out great") is fine. One identical reply pasted under fifty reviews does more harm than none.

5. Admin: invoices, scheduling notes, job summaries

This is where most contractors already use AI, and it's safe ground: a person sees every output, and mistakes are cheap to fix. Use the features already in your job-management software before paying for anything new.

AI calls and texts: the legal line

This is the part vendors skip, and it's the one that can cost real money.

Many "AI follow-up" products make outbound phone calls with a synthetic voice or send automated texts. In the US, the Telephone Consumer Protection Act (TCPA) governs both.

AI voices count as "artificial" voices. On 8 February 2024 the FCC issued a declaratory ruling confirming that AI-generated voices are "artificial" under the TCPA. Calls using them are covered by the same rules as prerecorded calls, which generally means you need the person's prior express consent (and prior express written consent for marketing calls) before the call.
The "one-to-one consent" rule never took effect. The FCC had planned to require consent for one seller at a time, which mattered for lead marketplaces. On 24 January 2025 the US Court of Appeals for the Eleventh Circuit vacated that rule in Insurance Marketing Coalition v. FCC. The underlying consent requirements still apply.

What that means in practice:

An AI calling back a homeowner who just asked you for a quote on your own website is a very different situation from an AI cold-calling a bought list.
Get clear consent for calls and texts on your forms and in your chat ("Can we call or text you about your project?"), and keep a record of it.
Ask any vendor selling AI calling: how does it record consent, how does it handle "stop", and who's liable if it gets it wrong?

We're engineers, not lawyers, and this isn't legal advice. Rules also vary by state. If outbound AI calls or bulk texting are part of the plan, talk to a lawyer who handles TCPA before you switch it on.

What to leave to a person

AI shouldn't be making these decisions in a contracting business:

Leave to a person

Why

The final price

A quote is a commitment. AI can give your published ranges and explain what moves the price. It shouldn't produce a number a customer can hold you to.

Site conditions

Soil, slope, access, structure, existing plumbing and electrics. A photo can help triage an enquiry, but it isn't a site visit.

Permits, codes and HOA rules

They vary by city and county and change over time. AI can say "we handle permits". It shouldn't tell a homeowner what their county requires.

Contracts, warranties and change orders

Anything that changes what you owe the customer.

Complaints and disputes

An upset customer with a leaking shower needs a person quickly, not a chatbot. Route these straight to someone.

A good AI tool makes the handover obvious. When a conversation reaches one of these topics, it should say a team member will follow up, collect the details, and stop.

When a custom build makes sense

Most contractors should buy, not build. Ready-made products cover intake, follow-up, quoting and admin well, for a monthly fee that's usually far less than custom development. We set out the numbers in what a RAG chatbot actually costs: build, buy, or have it built.

A custom system starts to pay off when:

Your pricing knowledge is unusual and valuable. For example, years of past jobs that could give homeowners a much better early estimate than a generic range, with a person still confirming every quote.
You run several brands, locations or franchises and need one system with separate data, rules and reporting for each.
Your tools don't talk to each other. Your CRM, estimating, scheduling and accounting systems need connecting, and no off-the-shelf product connects the ones you use.
AI is part of what you sell. A design-visualisation tool for your showroom, or a customer portal that explains project progress.

If you do commission a build, ask the vendor the questions in 12 questions to ask an AI development company before you sign. Two of them matter especially for contractors:

What happens when the model is retired? OpenAI has announced that GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini will be removed from its API on 14 January 2027. If a tool built for you in the last two years runs on one of them, someone has to move it to a newer model and re-test it before that date.
How does it stay up to date? If the assistant answers from your website, it has to be refreshed when your services and prices change. Otherwise last year's pool package price keeps being quoted.

A 90-day plan

When

Do this

Cost

Week 1

Check where website enquiries go, and how fast they're answered, including weekends

Free

Weeks 1–2

Publish price ranges and answers to common questions on your website

Your time

Weeks 2–4

Add an AI website assistant for after-hours questions and qualification

Monthly subscription

Month 2

Use AI to draft proposals from your own templates and price book; a person checks every one

Usually included in your software

Month 2

Set follow-up reminders on every open quote; AI drafts, a person sends

Free to low

Month 3

Review: how many leads came in after hours, how fast they were answered, how many became site visits

Free

Later

Consider a custom build only if one of the four conditions above applies

Quote-based

Measure before and after. The point isn't to "use AI". It's to answer more enquiries faster and sign more of the jobs you already quote.

Frequently asked questions

What is the best use of AI for a contractor?

For most remodelers and pool builders, answering and qualifying website enquiries, especially in the evening and at weekends when nobody's in the office. After that, drafting proposals and follow-up messages, with a person checking each one.

Can AI give my customers a price?

It can give your published price ranges and explain what affects the price. It shouldn't give a final quote. A quote is a commitment, and it depends on site conditions an AI can't see. Set a rule like "never quote a final price, offer a site visit".

Are AI phone calls to customers legal?

In the US, AI-generated voices count as "artificial" voices under the TCPA (FCC ruling, February 2024), so calls using them generally need the person's prior express consent, and prior express written consent for marketing calls. Calling back someone who asked you for a quote is different from cold-calling a list. Check with a lawyer before using AI for outbound calls or bulk texts.

How many contractors use AI?

In Jobber's December 2025 survey of 1,050 US home service business owners, 52% used AI in day-to-day operations, mostly for quoting (54% of users), invoicing (52%) and business writing (51%). Among general contractors it was 57%.

Should a contractor build a custom AI tool?

Usually not. Ready-made tools cover intake, follow-up and proposals well. A custom build makes sense when your pricing data is unusually valuable, you run several brands or locations, your systems need connecting in ways no product supports, or AI is part of what you sell.

Sources

Jobber, 2026 Home Service Trends report, survey of 1,050 US home service business owners via Conjointly, December 2025, ±3 points at 90% confidence.
BrightLocal, Local Consumer Review Survey 2026, published 11 February 2026; 1,002 US adults via SurveyMonkey.
US Federal Communications Commission, declaratory ruling on AI-generated voices in robocalls, 8 February 2024.
Morrison Foerster, Eleventh Circuit vacates FCC's TCPA one-to-one consent rule, January 2025, on Insurance Marketing Coalition v. FCC, decided 24 January 2025.
OpenAI, API deprecations, notice of 14 October 2026.

Related reading

Brillnex Systems builds AI agents, assistants and integrations for businesses, and runs its own AI product in production. If one of the four custom-build conditions above fits your business, tell us what you're trying to fix, or see our AI agent development services.

AI for Home Improvement Contractors in 2026: What to Automate First, and What to Leave to a Person