Two years ago, “AI for contractors” was a slide in every vendor pitch deck and almost nothing in production. As of 2026, that has flipped. ServiceTitan’s 2026 In the Trades report (1,000+ contractors surveyed) shows AI adoption with measurable business impact jumped from 17% in 2025 to 38% in 2026. That’s not noise. That’s a real curve.
Below is what’s actually working in home improvement sales right now, with the platforms, the numbers, and the failures. No vaporware.
Where AI is genuinely moving the needle in 2026
1. Voice AI for in-home sales coaching (Rilla, others)
This is the single most-talked-about AI use case in residential home services this year. Tools like Rilla record and transcribe in-home sales conversations, then use AI to coach reps on talk-to-listen ratio, objection handling, and closing language. Used heavily across HVAC, plumbing, roofing, and remodeling.
The case study everyone in the industry is trading: Preferred Home Services, where one sales associate moved from a 50% close rate to 71% in two months using Rilla, and became the first plumbing sales associate at the company to clear $1M in annual sales. Managers do up to 10 virtual ridealongs per day vs. 10 per week before. A to Z Dependable Services reports a new rep hit 50% close rate in his first month, average ticket up about 30%, and managers compressing two days of ridealongs into 30 minutes.
Rilla’s framing: managers can coach “8x faster and 20x more efficiently.” The math is real because you’re no longer sitting in a truck for 8 hours to observe one conversation.
2. AI lead response and CSR (Hatch, ServiceTitan AI)
Speed-to-lead is the closing variable nobody on your team is winning at. Industry data is consistent: leads contacted within 5 minutes convert at roughly 9x higher rates than leads contacted 30+ minutes later. About 79% of marketing leads never convert primarily because follow-up is slow or inconsistent.
Hatch (acquired by Yelp in January 2026 for around $270M) runs AI agents that text, call, and email inbound leads to book appointments. ServiceTitan’s AI Voice Agents and SMS Booking Agent handle call overflow on the same model. Voice AI on inbound calls is showing roughly 60% higher conversion vs. unanswered or late-callback paths.
Where it pays off: shops that were missing 30 to 60% of inbound calls during business hours and 100% of after-hours.
3. AI lead scoring and follow-up inside the CRM (JobNimbus)
JobNimbus uses ML to flag leads most likely to close and auto-runs text and email cadences. ServiceTitan’s residential CRM (general availability spring 2026) adds a centralized speed-to-lead follow-up queue that pings reps the moment a lead lands.
The honest take: lead scoring at low volume is a parlor trick. Below 200 leads a month, the model doesn’t have enough signal to be smarter than your gut. Above that, it starts to earn its place.
4. Voice-to-CRM data entry (JobNimbus Scout)
JobNimbus announced “Scout” in January 2026 (closed beta), letting reps create jobs and update statuses by voice from the truck. The win here isn’t the AI itself. It’s adoption. Your reps were never going to type up structured notes after every appointment, but they will talk into their phone for 90 seconds. That 90 seconds becomes a clean CRM record automatically.
5. AI estimating from photos and aerial imagery (Togal.AI, SumoQuote)
Togal.AI, JobNimbus AI-assisted estimates, and SumoQuote pull measurements and generate proposals in minutes. Coastal Construction reported roughly $1M in savings using Togal.AI through faster, more accurate estimating.
The caveat: aerial AI on complex roofs (tree cover, snow, dormers, multi-pitch, skylights) still produces silent errors. Missing two squares on a 40-square roof is a four-figure mistake. Use AI for first drafts, human for final numbers.
6. Photo-to-report automation (CompanyCam AI Notes)
CompanyCam AI Notes turns jobsite photos into inspection reports with auto GPS and timestamps. Mostly useful for storm/insurance work and warranty claims. Not transformative on its own; meaningful when stacked on top of the rest of the workflow.
The numbers that matter
- 38% of contractors report measurable AI business impact in 2026, up from 17% in 2025 (ServiceTitan).
- 74% of residential contractors see AI as key to efficiency.
- AI adoption by function: administrative 59%, marketing & sales 51%, cost estimation 24%, bid management 22%.
- Only 10% of contractors view AI as a competitive advantage. The other 64% see it as efficiency. That gap is where the early movers are eating.
- Top adoption barriers: training and integration complexity (44%), unclear ROI (37%), employee resistance (18%).
- Average home services cost per lead in 2025: $90.92; Google roofing leads $187.79. AI lead recovery is paying for itself before you finish the trial.
Where AI is still failing contractors
Standalone aerial measurements on complex roofs
Models output a number even when they miss skylights, dormers, or shaded sections. Verify against ground measurements on anything non-rectangular.
Full automation of high-ticket closes
Scope Technologies, in their post on AI failures: “AI alone can’t price or close high-ticket projects, or protect margins, without a human safety net.” A $30K bath remodel still closes at the kitchen table.
Pilots that never scale
RICS’s 2025 global construction report found nearly half of construction firms have no AI in production and roughly a third are stuck in proofs of concept. The pattern is the same across home improvement.
Trust and data infrastructure
ServiceTitan’s data shows about 25% of residential contractors use AI meaningfully and nearly 50% don’t trust it yet. The other half-truth: most contractors lack the clean, connected job and customer data AI needs to actually work.
Replacing humans on emotional or complex calls
An AI voice agent will not handle a no-heat call from a panicked homeowner with three kids in winter. It will not negotiate scope on a $50K kitchen redesign. The “AI replaces salespeople” pitch is wrong. The right framing is AI removes the parts of selling that aren’t selling.
How to test an AI tool in 7 days without risking your business
- Pick one workflow. Voice coaching, lead response, voice-to-notes, AI estimating. One.
- Run it in parallel for 7 days, not as a replacement. Your human keeps doing what they do. The AI runs alongside.
- Compare outputs. Did the AI version save time? Was the quality acceptable? Did anything embarrassing make it through?
- Then commit or kill. If it earned its place in 7 days, expand. If not, kill it. AI is too fast-moving to commit to anything that doesn’t pay back quickly.
The compounding stack: AI + CRM + financing
The biggest mistake we see is contractors treating AI as a standalone tool. AI is multiplication, not addition. An AI follow-up sequence that says “hi, just checking in” is noise. An AI follow-up sequence that says “hi, I noticed you were pre-qualified for $312/month financing on the bath remodel, that approval expires Friday, want to lock the schedule?” is a closing tool. Same AI. Different stack underneath it.
If you’re stacking, the order is: clean CRM data first, financing options second, AI on top of both. Skip the foundation and the AI will look like it isn’t working.
The honest take
AI in home improvement sales in 2026 is not hype anymore in the use cases above. The Rilla close-rate lifts are documented. Hatch (now Yelp) didn’t get bought for $270M on a slide deck. Speed-to-lead AI agents are recovering deals your team was losing yesterday.
What’s still hype: full autonomy, AI replacing salespeople, AI lead scoring at low volume, AI estimating without human review. Anything that promises to remove the human entirely from a high-trust transaction is overselling.
Pick one workflow. Run it for a week. Then expand.