What is the single best-evidenced reason to distrust a home-services AI vendor's performance claims?

This exact buyer population has already been burned once by an unaudited vendor claim, and a federal regulator wrote down what it found. HomeAdvisor, Inc. — a lead-generation marketplace serving the same HVAC, plumbing, electrical and general home-improvement contractors this article is about — was the subject of FTC Matter No. 1923106, Docket No. 9407. The Commission's March 2022 administrative complaint alleged HomeAdvisor made false, misleading or unsubstantiated claims about the quality and source of the leads it sold, told contractors its leads converted to paying jobs at rates it could not substantiate, and misrepresented an optional add-on subscription as free. In January 2023 the FTC ordered up to $7.2 million in payments to affected contractors, split into two redress funds — up to $3.2 million at $30 per membership and up to $4 million at $59.99 per mHelpDesk membership — and approved the final order in April 2023, with over $3 million actually returned by November 2023. That case is about a lead marketplace, not an AI booking agent, and we do not conflate the two. But it establishes something directly relevant: nobody has independently audited an AI phone or dispatch agent's booking-rate claim in this vertical, and the one time a regulator audited an adjacent vendor's performance claim to this same buyer population, it found the claim could not be substantiated.

Is there an independent benchmark of AI phone or dispatch agents for HVAC, plumbing or electrical companies?

No, and we looked. Restaurants have one credible third-party measurement of drive-thru voice AI (Intouch Insight's mystery-shop study, published with QSR Magazine). We could not find an equivalent independent study — no mystery-shop research, no academic paper, no trade-association benchmark — of any commercial AI phone-answering, dispatch or booking product serving home service contractors, checked as of 2 September 2026. Every accuracy, booking-rate or automation-rate figure you will encounter in this category is self-published by the vendor selling the product. That absence is itself the finding: it means every ranking of this category, including this one, is working from vendor marketing on the performance question, which is exactly why this article does not score performance at all.

Does the FTC's Cooling-Off Rule apply to a $9,000 HVAC replacement quote signed in someone's kitchen?

Yes, and the rule does not care whether AI helped generate the number. The FTC's Cooling-Off Rule, 16 CFR Part 429, makes it an unfair and deceptive practice for a seller to complete a door-to-door — meaning in-home or at a location other than the seller's permanent place of business — sale worth more than $25 without giving the buyer written notice of a three-business-day right to cancel, disclosed both orally and in a completed “NOTICE OF CANCELLATION” form. If a technician or sales rep sits at the kitchen table and closes an $9,000 system-replacement agreement, that transaction is squarely inside the rule regardless of whether an AI tool priced the job, coached the rep's pitch, or drafted the paperwork. The obligation runs to the seller, not to the software vendor.

Does California require an AI phone agent to disclose that it's a bot before it tries to book a job?

Yes, when the interaction is meant to incentivize a purchase from a California consumer. California's Bot Disclosure Law, enacted as SB 1001 and codified at Business and Professions Code § 17941, has made it unlawful since it became operative on 1 July 2019 to use a bot to communicate online with a person with the intent to mislead them about its artificial identity for the purpose of incentivizing a sale, unless the bot's operator discloses — clearly, conspicuously, and in a manner reasonably designed to inform — that it is a bot. A bill that would have tightened this considerably, AB 410, would require the bot to disclose its nature at first contact and answer truthfully if asked whether it is human; as of the most recent legislative record we could verify, the bill was in an engrossed, held-under-submission status as of 29 August 2025 and had not been signed into law. Treat the current, operative law as § 17941 as written, and check AB 410's status directly before assuming it is in force.

Is travel time between job sites during the workday paid, or does it count as an unpaid commute?

It is paid, and this is one of the most commonly misapplied wage-hour rules in the trades. Under the Fair Labor Standards Act, the U.S. Department of Labor draws a firm line: home-to-first-job travel and last-job-to-home travel are ordinarily unpaid commuting, but any travel between job sites during the workday is compensable hours worked — the Department's own guidance gives the example of a plumber who makes house calls, whose customer-to-customer travel time counts. If a technician is required to stop at the shop first to pick up parts or get dispatch instructions, the shop-to-first-job leg is compensable too. A routing or dispatch agent that quietly restructures a technician's day — more stops, tighter windows, different sequencing — can shift how much of that day is compensable travel without anyone intending it to, which makes this a payroll question dressed as a scheduling optimization.

Can an AI dispatch agent legally assign any available technician to any job on the board?

No, not if the job requires a licensed trade classification the technician does not hold. Electrical work, in most states, requires a licensed electrician or a supervised apprentice under specific ratios; plumbing and HVAC carry parallel licensing regimes that vary by state and sometimes by municipality. Contracting without the required license carries real consequences — California's Contractors State License Board, as one illustrative example, can pursue unlicensed contracting as a misdemeanor carrying up to six months in jail and a $5,000 fine, on top of a separate administrative fine ranging from $200 to $15,000. That is one state's numbers, offered as an illustration, not a national figure — your state's exposure will differ, and this is a question for local counsel, not a settled national rule. The practical point for a dispatch agent is that license-class matching has to be enforced in the assignment logic itself, not left to a technician to self-correct in the field.

Which home-services AI vendors in this roster publish real, dollar-amount pricing?

Four of the eleven, as checked on 2 September 2026. Housecall Pro publishes Basic at $79 per month, Essentials at $189, and MAX at $329 (each lower on annual billing), with extra MAX users at $35 per month. Jobber publishes solo pricing from $29 per month up to team pricing near $529 per month for its largest tier. Goodcall publishes three tiers — $79, $129 and $249 per month — each with a unique-caller cap and a $0.50 per-caller overage beyond it. CallRail publishes four annual-billing tiers from $50 to $195 per month, though its AI Conversation Intelligence feature is gated behind the $150-and-up tier. The other seven — ServiceTitan, Weave, FieldEdge, Invoca, Broadly, Rilla and Podium — either gate pricing behind a sales conversation or, in Podium's case, have pricing reported only by third-party review sites that we could not independently confirm against the vendor's own page this session.

Do any of these vendors disclose what share of calls or bookings their AI completes with no human involved?

None that we could find. We checked every vendor's own site in this roster on 2 September 2026 and found no published human-in-the-loop rate — no vendor states what percentage of calls, bookings or dispatches its AI completes without a CSR, dispatcher or the technician themselves stepping in to correct or complete the interaction. Given that the one adjacent enforcement action in this buyer population (the FTC's HomeAdvisor order) turned on exactly this kind of undisclosed gap between marketed performance and actual human involvement, this is the single most valuable written question to put in an RFP, and the easiest one for a vendor to dodge with a vague answer.

What happened to Weave, and why does its status matter for a home-services buyer?

Weave Communications, publicly traded on the NYSE under WEAV, agreed to be acquired by private equity firm Francisco Partners for approximately $650 million, or $7.40 per share — roughly a 34% premium to its trading price — with the board's approval and a close expected in the fourth quarter of 2026, subject to shareholder and regulatory approval. As of the deal's announcement it had not closed. Separately, and this matters for fit: Weave's core, largest customer base is dental, optometry, medical and veterinary practices, serving over 40,000 locations; it also publishes a dedicated HVAC industry page, so it does market to the trades, but a buyer should go in clear-eyed that this is a healthcare-first communications platform extending into home services, not a home-services-native product.

Is Rilla a phone-answering or booking agent like the others in this roster?

No, and grouping it with the phone agents would misdescribe it. Rilla records, transcribes and analyzes in-person sales conversations — the pitch a technician or sales rep gives standing in a homeowner's kitchen or driveway, most often for big-ticket replacement or installation work — and coaches the rep against that recording. It is not a call-answering or job-booking product. That makes it the vendor in this roster most directly implicated by the Cooling-Off Rule discussion elsewhere in this article: the conversation Rilla is built to record and coach is frequently the same conversation that triggers a $25-plus in-home sale's three-day right-to-cancel disclosure obligation, and in two-party-consent states, the recording itself raises a separate consent question independent of anything to do with AI.

Should an AI agent ever quote a firm price for a system replacement or installation over the phone?

Treat a flat-rate diagnostic or repair line item from a published pricebook — the kind of rate a platform like FieldEdge's Coolfront pricebook is built to hold — differently from a full system replacement or installation, which usually requires an in-home assessment before a firm number is honest. An agent that reads back a documented flat rate for a known, bounded repair is on solid ground. An agent that commits to a $9,000 installation price sight-unseen is making a representation a licensed human hasn't verified, and if that number becomes a signed in-home agreement, the Cooling-Off Rule disclosure obligation attaches regardless of who or what produced the figure.

What is the honest cost and timeline for a home-services agent build?

Frenchy Digital's bands are discovery plus workflow audit at $9k–$22k over 2–4 weeks; a single-workflow agent at $28k–$70k over 4–9 weeks; a multi-workflow platform with system integration at $70k–$180k over 9–16 weeks; and an enterprise, multi-location or regulated build at $180k–$420k+ over 14–24 weeks. Senior-led work runs $150–$225 per hour, retainers run $2,500–$9,500 per month, every build carries a 30-day post-launch warranty, and full source-code and IP ownership transfers to you. We return a fixed-price phased proposal within 5 business days of a discovery call. If a published-price product — a $79-a-month AI phone agent, say — already covers your workflow, buying it is usually the right call, and we will tell you so rather than propose a custom build you do not need.

Are “missed call revenue” calculators that vendors post on their pricing pages trustworthy?

No, and we do not print their outputs. Several vendors in this category, and in adjacent categories we have reviewed on this site, ship a “missed call revenue calculator” or an equivalent unsourced dollar figure claiming what an average contractor loses to unanswered calls. None we have found discloses a sample, a methodology, or a source — it is a marketing input dressed as a statistic. The number that actually matters is free and already sitting in your own phone system or call-tracking log: pull the last 90 days of call records, count the calls that went unanswered, and check how many of your booked jobs in that window came from a first call versus a callback. That is a real number about your business. A vendor's national average is not.

What should go in a home-services company's RFP to an AI phone or dispatch vendor?

Five written questions do most of the work. First, what percentage of calls or bookings in the last 90 days at comparable customer sites were completed with no human — CSR, dispatcher or technician — involved, and how is that measured? Second, how does the product's bot-disclosure behavior change by state, and specifically how does it comply with California's Bus. & Prof. Code § 17941 for California callers? Third, how does the dispatch logic enforce license-class matching, and can it produce an audit trail showing which license a given assignment relied on? Fourth, what is the data retention and deletion schedule for call recordings and transcripts, and does the contract grant the vendor rights to use your customers' data to train its own models? Fifth, is pricing genuinely fixed at the published tier, or does it change once your call volume or location count crosses a threshold the salesperson didn't mention up front?

Back to Blog

AI Agents

September 2, 2026

30 min read

Top 10 AI Agents forHome Service Contractors in 2026

Every ranking of home-services AI you'll read scores accuracy, booking rate or ROI. Every one of those figures is published by the vendor about itself, and nobody has independently audited a single one in this vertical. So this ranking scores only what an owner can re-check from a browser and a contract — and says out loud what it refused to score, including the closest thing this exact buyer population has to a cautionary tale: an FTC order against a lead-generation vendor that made claims nobody could substantiate.

AI agents for HVAC, plumbing and electrical contractors in 2026 — phone answering, dispatch and sales-coaching agents ranked on verifiable attributes

$7.2 million

Up-to redress amount the FTC ordered HomeAdvisor to pay contractors over unsubstantiated lead-quality claims — the one adjacent instance of a regulator auditing a vendor's performance claims to this buyer population

Federal Trade Commission, In the Matter of HomeAdvisor, Inc., File No. 1923106, Docket No. 9407 — order January 2023, final order approved April 2023

$625 million

Raised in ServiceTitan's IPO, the only publicly traded, SEC-reporting company in this roster

ServiceTitan, Inc. (Nasdaq: TTAN) initial public offering, priced 11 December 2024, began trading 12 December 2024

~1.76 million

Combined US employment across electricians, plumbers/pipefitters/steamfitters, and HVAC mechanics and installers — the workforce this entire roster has to fit into

US Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025 release — 757,220 + 465,840 + 541,070, summed by us from the three published occupation totals

$650 million

Agreed buyout value for Weave Communications' pending take-private by Francisco Partners at a roughly 34% premium — expected to close Q4 2026, not yet closed

Weave Communications, Inc. (NYSE: WEAV) merger agreement disclosure, 2026

Key Takeaways

Every ranking of home-services AI you'll read scores accuracy, booking rate or ROI. Every such figure is published by the vendor about itself. Nobody has independently audited one, so this ranking scores only what an owner can re-check: pricing, integrations, compliance and ownership.
The closest thing to a regulator auditing a vendor's performance claim to this buyer population is the FTC's order against HomeAdvisor — File No. 1923106, Docket No. 9407 — which found the lead marketplace's quality and conversion claims unsubstantiated and ordered up to $7.2 million in redress. It is about leads, not AI, and we do not conflate the two, but it is the clearest evidence this exact audience has been burned by an unverifiable vendor claim before.
No independent benchmark of AI phone, dispatch or booking agents for this vertical exists — unlike restaurants, which have one mystery-shop study. Every accuracy or automation figure in this category is self-reported.
The binding constraint most owners underrate is the FTC's Cooling-Off Rule, 16 CFR Part 429: any in-home sale over $25 carries a mandatory three-business-day right to cancel, regardless of whether AI priced, pitched or paperworked the deal.
California's Bot Disclosure Law (SB 1001, Bus. & Prof. Code § 17941) has required bots that try to incentivize a sale to disclose their artificial identity since 1 July 2019. A stronger amendment, AB 410, was held under submission in committee as of the most recent record we could verify — check its status before assuming it is law.
Only four of eleven vendors publish real pricing: Housecall Pro, Jobber, Goodcall and CallRail. Every other cell in the roster reads 'not publicly disclosed' or 'not verified this session' — and we never estimate on a vendor's behalf.
No vendor in this roster discloses a human-in-the-loop rate for calls or bookings its AI completes unassisted. Given the HomeAdvisor precedent, that is the single most valuable written question for an RFP.
Weave (NYSE: WEAV) agreed to a ~$650 million take-private by Francisco Partners at a ~34% premium, expected to close Q4 2026 — not yet closed. Rilla is not a phone or booking agent at all; it is in-person sales-conversation coaching, which is exactly the conversation the Cooling-Off Rule governs.
Travel time between job sites during the workday is compensable under the FLSA — a dispatch or routing agent that restructures a technician's day can quietly create wage-hour exposure nobody scoped for.
Frenchy Digital bands: discovery and workflow audit $9k–$22k; single-workflow agent $28k–$70k; multi-workflow platform with system integration $70k–$180k; enterprise, multi-location or regulated build $180k–$420k+.

The Claim Under Test

Every listicle ranking AI tools for HVAC, plumbing and electrical contractors you will read this year is built on numbers nobody outside the vendor has checked. Open a handful and the pattern repeats: a phone agent claims it books 40% more jobs, a dispatch tool claims it cuts drive time by a third, a lead-response product claims a specific dollar figure in “recovered revenue.” None of those figures comes with a sample size, a denominator, or an independent auditor. They are marketing assertions, printed as though they were measurements.

This buyer population has already been burned by exactly this pattern once, and it is on the public record. HomeAdvisor — a pay-per-lead marketplace serving the same contractors this article is written for — was the subject of an FTC administrative complaint alleging its lead-quality and conversion claims could not be substantiated. The Commission ordered up to $7.2 million in redress. That case is about a lead marketplace, not an AI phone or dispatch agent, and this article does not conflate the two. But it means the closest thing to a regulator auditing a vendor's performance claim to this exact audience already happened, and the claim did not hold up.

So the claim under test here is narrow: can a contractor rely on any published performance figure in this category? Having checked every vendor site in this roster on 2 September 2026, our answer is no — and the useful response is to stop trying. Rank on what you can verify from a browser and a contract, put performance questions into your RFP as written commitments rather than homepage numbers, and measure your own call and booking data at your own locations.

The two facts that carry this article.First: no independent benchmark of AI phone, dispatch or booking agents for home service contractors exists — every accuracy or automation figure you'll see is self-published. Second: the one adjacent regulatory finding in this exact buyer population, the FTC's HomeAdvisor order, found a vendor's performance claims to contractors could not be substantiated. Different product category, same lesson: verify before you buy, and don't expect anyone else to have verified it for you.

This piece sits inside a wider set of vertical rankings built on the same method; the cross-industry version is our guide to the top AI agents in 2026. What follows is the home-services edition: what we scored and refused to score, eleven vendors in order, the constraints that decide whether any of it works safely, and the parts we could not verify.

How We Ranked, and What We Refused to Rank On

We scored seven attributes, every one re-checkable by a reader in under an hour.That constraint is the whole method: if you can't independently confirm a cell in our table, it belongs in a sales deck, not a ranking.

What we scored — every attribute re-checkable by the reader

Documented public integrations — named on the vendor's own site, not a partner's
Pricing transparency — published dollar amounts versus a gated sales conversation
Whether the product is standalone or locked to a suite
Ownership and corporate status from the public record — SEC filings, completed acquisitions, press releases
Whether any independent evaluation of the product exists
Whether the vendor discloses a human-in-the-loop rate for calls or bookings
Fit to this specific vertical — home services trades — versus a general-purpose or adjacent-vertical product wearing a home-services page

We refused to score accuracy, booking rate, automation rate, calls saved, revenue recovered and ROI.Not because they don't matter — they're the only things that matter operationally — but because every published figure in this market is self-reported by the seller, with no neutral party checking any of it, and no shared definition of what any of these terms even means from one vendor to the next.

The claim	Where it comes from	What we print instead
“You're losing $X a year to missed calls” / missed-call revenue calculators	Vendor lead-magnet tools with no disclosed sample or methodology — several vendors in this category ship a version of this on their own pricing pages	Pull your own last-90-days call log and count actual missed calls against actual booked-job value. That number is free, and it is about your business
“Homeowners hire the first contractor who calls them back”	Recycled marketing copy with no traceable original study	Nothing. We could not find a primary source for this claim in any form we checked, so we do not repeat it
Any vendor's self-published booking rate, automation rate, answer rate or ROI figure	Self-reported by the seller, with no independent audit of any product in this category	The FTC's finding about a lead-quality claim in this exact buyer population, and a written RFP question demanding the same figure with methodology attached
“Our leads convert at rates that will grow your business” (or similar lead-quality assurances)	The marketing language the FTC's 2022 administrative complaint against HomeAdvisor specifically challenged	The FTC's own findings and order, with the redress-fund structure that followed them
A single national dollar figure for unlicensed-contracting exposure	State licensing penalties vary by jurisdiction; no single number represents fifty states	One state's own published range (California's CSLB) as an illustration, with an explicit note that your state differs and needs local counsel

Methodology block — how to re-check every cell yourself. Date checked: 2 September 2026. Method: we opened each vendor's own pricing, product and integration pages directly; we pulled corporate status from SEC filings, completed-transaction press releases and named investor disclosures; we pulled statutory text from the FTC, the eCFR, the FCC and the California Legislature's own sites; we pulled the DOL travel-time rule from the Department's own guidance. Where a claim came only from a third-party review site rather than the vendor's own page, we say so and do not present it as vendor-confirmed. We never estimate a price. To re-check: open the vendor URL in our sources list, look for a published number, and note the date you looked. A different result than ours means the vendor changed the page — itself worth knowing.

What a Regulator Found the One Time It Audited This Industry's Claims

No US regulator has audited an AI phone or dispatch agent's performance claim to home service contractors.But one has audited the closest adjacent product this exact buyer population relies on — a pay-per-lead marketplace — and the result is instructive for anyone evaluating a vendor's booking-rate claim today.

The case is In the Matter of HomeAdvisor, Inc., FTC File No. 1923106, Docket No. 9407. The Commission's administrative complaint, filed in March 2022, alleged that since at least mid-2014 HomeAdvisor made false, misleading or unsubstantiated claims about the quality and source of the home-improvement leads it sold to service providers — including that leads matched a provider's stated services and geography, and that leads converted into paying jobs at rates the company could not substantiate. The complaint separately alleged that sales agents misrepresented an optional one-month mHelpDesk subscription as free.

FTC Order Requires HomeAdvisor to Pay Up To $7.2 Million and Stop Deceptively Marketing its Leads for Home Improvement Projects.
— Federal Trade Commission, press release, January 2023

The order's structure is worth reading closely, because it shows how these redress calculations actually work. Up to $3.2 million was set aside for payments of up to $30 per membership tied to the lead-quality claims, and up to $4 million for payments of up to $59.99 per mHelpDesk membership tied to the “free” misrepresentation. The FTC approved the final consent order in April 2023, and by November 2023 had returned more than $3 million directly to affected contractors, with a claims process opened for additional refunds.

What this case is not, and getting that right matters.It is not an AI-washing case — there is no allegation that HomeAdvisor misrepresented an AI capability, because the product at issue is a lead marketplace, not an AI agent. It does not find that any of the eleven vendors ranked in this article made similar claims; none of them is a party to it. What it establishes is narrower and still useful: the exact population of buyers this article is written for has, at least once, paid for a vendor's unverified performance claim and later needed a federal order to get money back for it.

Set that alongside the second finding of this section: no independent benchmark of any AI phone, dispatch or booking product for home service contractors exists.We looked for a mystery-shop study, an academic evaluation, or a trade-association benchmark comparable to what restaurants have in the Intouch Insight drive-thru study, and found none, checked as of 2 September 2026. Every accuracy, booking-rate or automation figure you will see attached to a product in this roster is self-reported by the company selling it, with no external party checking the math. Combine the two findings and the practical conclusion is the same one that opened this article: verify what you can, and don't buy a number nobody but the seller has ever measured.

The Comparison Table

Every cell below is either a citation you can open or the literal words “not publicly disclosed” or “not verified this session.”There is no accuracy column, no booking-rate column and no star rating, because none of those could be filled honestly. All entries were checked on 2 September 2026 against the vendor's own pages and the public corporate record.

Vendor (checked 2026-09-02)	What it actually is	Published pricing	Named integrations on the vendor's own site	Ownership of record
1. ServiceTitan	FSM platform with an in-house Pro Products suite — Scheduling Pro, Dispatch Pro, Contact Center Pro with named AI Voice Agents	Not publicly disclosed	Contact Center Pro's Universal Inbox consolidates calls, texts, emails and bookings across locations; no third-party integration names published on the Pro Products pages we checked	Public — Nasdaq: TTAN, IPO priced 11 December 2024, raised ~$625 million
2. Housecall Pro	All-in-one FSM with an in-house AI Team suite — Marketing AI, CSR AI Chat Answering, Analyst AI, Coach AI, Help AI	Yes: Basic $79/mo, Essentials $189/mo, MAX $329/mo (billed monthly; lower annually); extra MAX users $35/mo	QuickBooks referenced in third-party reviews; not enumerated on the pricing page we checked	Private — investor detail not verified this session
3. Weave	Patient/customer communications platform with AI-driven engagement; publishes a dedicated HVAC industry page	Not publicly disclosed	Primarily positioned for dental, optometry, medical and veterinary practices; the HVAC page does not enumerate FSM integrations	Public — NYSE: WEAV; agreed to a ~$650 million take-private by Francisco Partners (~$7.40/share, ~34% premium), expected close Q4 2026, not yet closed
4. Jobber	FSM software with an add-on AI Receptionist	Yes: solo pricing from $29/mo (Core) up to team pricing near $529/mo (Plus, 15 users); Receptionist add-on price conflicts across sources ($29 vs $99/mo) and is included free on Plus	QuickBooks integration referenced in vendor materials; full list not verified this session	Private — Edmonton, Canada
5. Goodcall	Standalone AI voice agent for inbound calls, configured from a business's Google Business Profile	Yes: Starter $79/mo (100 unique customers), Growth $129/mo (250), Scale $249/mo (500), unlimited minutes, $0.50 per customer over the cap	Documented integration with FieldEdge; others not verified this session	Private
6. FieldEdge (by Xplor)	FSM software for HVAC, plumbing and electrical — infrastructure, not an agent; no native AI dispatch, receptionist or booking as of the sources we checked	Not publicly disclosed	Two-way QuickBooks Desktop sync; Coolfront flat-rate pricebook; integrates with Goodcall as a third-party AI phone assistant	Private — Xplor Technologies, backed by Advent International, Battery Ventures and Silver Lake
7. CallRail	Call tracking platform with AI Conversation Intelligence gated behind its higher tiers	Yes: Lead Tracking $50/mo, Lead Tracking Complete $95/mo, Lead Conversion $150/mo, Lead Conversion Complete $195/mo (annual billing); AI features require Lead Conversion or above	Broad marketing-stack integrations referenced in third-party reviews; not enumerated on the pricing page we checked	Private — majority institutional ownership (Sageview Capital, Goldman Sachs Asset Management) reported as of 2024
8. Invoca	Enterprise conversation intelligence and call tracking across many verticals, including home services	Not publicly disclosed; 30-day free trial offered	Not enumerated on the pages we checked	Private — venture-backed (Accel, Upfront Ventures, Silver Lake Waterman, H.I.G. Growth Partners, Salesforce Ventures); ~$184 million raised
9. Broadly	Reputation, messaging and payments hub with AI-timed review requests, marketed heavily to home-services SMBs	Not publicly disclosed	Not enumerated on the pages we checked	Acquired by Vendasta, announced 17 March 2023; Vendasta is privately held
10. Rilla	AI conversation intelligence and sales coaching for in-person field sales reps — not a phone or booking agent	Not publicly disclosed; a third-party research estimate implies an average contract value near $20k/year, not vendor-confirmed	Home Depot partnership announced January 2026; Shore Consulting sales-methodology partnership from July 2025	Private — raised a $68.5 million Series B; 2025 private-market valuation estimated at $737–758 million by third-party data providers, not vendor-confirmed
11. Podium	Messaging-first AI Employee for lead response, reviews and payments	Reported by third-party review sites at Core $399/mo, Pro $599/mo, Enterprise $999+/mo, plus a $99/mo AI receptionist add-on — not independently confirmed against Podium's own pricing page this session	Not enumerated in the sources we could check this session	Private — last confirmed valuation $3 billion (November 2021); ~$425 million total raised (Accel, GV, Summit Partners, Y Combinator, Sapphire Ventures, IVP, Album VC, Kickstart Seed Fund)

Three patterns are worth pulling out on their own. First, published pricing is the exception, not the rule — only four of eleven vendors publish real dollar amounts, and even one of those four (Jobber's AI Receptionist add-on) has a genuinely unresolved discrepancy between sources rather than a clean confirmed number. Second, compliance documentation is nearly invisible in this category — we could not independently verify a SOC 2, PCI DSS or comparable certification for any of the eleven from the pages we checked this session, which is worth a direct question in any RFP given how much call and customer data these products touch. Third, and most important, not one vendor discloses a human-in-the-loop rate — nobody publishes what share of calls, chats or bookings its AI completes with zero human involvement. After the HomeAdvisor precedent, that is the single easiest, highest-value written question a buyer can ask, and the one every vendor in this roster currently leaves unanswered.

The Eleven, in Order

The order below reflects verifiability, not performance. A vendor that publishes its price, names its integrations and has a clear ownership record ranks above one that publishes nothing — not because its product works better, but because you can make an informed decision about it and hold it to what it said.

1. ServiceTitan — the most publicly disclosed company in the roster, on everything except price

What it does: a full field-service-management platform for the trades, with an in-house suite of named AI products — Scheduling Pro, Dispatch Pro, and Contact Center Pro, which bundles a Universal Inbox consolidating calls, texts, emails and online bookings across locations, a Manager Assist tool, and AI Voice Agents the vendor describes as able to recognize returning customers, check real-time capacity, and book jobs instantly.
What is verifiable: more corporate disclosure than any private vendor here can offer, because it is a public company. Nasdaq: TTAN, IPO priced at $71 per share on 11 December 2024, trading began the next day, and the offering raised approximately $625 million.
What is not disclosed:pricing, on either the base platform or the Pro Products suite — both are gated behind a sales conversation. We located no compliance trust page and do not assert a certification. Third-party reseller estimates circulating online (roughly $245–$1,200 per month depending on company size) are not ServiceTitan's own published numbers, so we do not repeat them as fact.
Who it fits: multi-location trades businesses already considering a full platform switch, for whom native AI features across scheduling, dispatch and the contact center matter more than a la carte pricing. Who it does not: a single-truck operator who wants one narrow workflow solved cheaply — that buyer is better served lower in this list.

2. Housecall Pro — the clearest published pricing and the broadest named AI suite

What it does:an all-in-one FSM platform with an AI Team built out across 2024–2026: Marketing AI drafts campaign copy, CSR AI Chat Answering handles inbound website chat and captures job details, Analyst AI answers natural-language questions about the business's own data, Coach AI gives growth and hiring guidance, and Help AI answers in-app platform questions — all five included on every plan.
What is verifiable: published pricing to the dollar. Basic at $79/month (one user), Essentials at $189/month (up to five users), and MAX at $329/month (up to eight users), each discounted on annual billing, with additional MAX users at $35/month and a 14-day free trial.
What is not disclosed:a live-voice CSR AI call-answering feature exists as a paid add-on alongside the included chat-only version, and we could not confirm its exact price this session. Compliance documentation and the company's current investor cap table were not verified this session.
Who it fits: small-to-midsize trades businesses that want a known monthly cost and a broad set of included AI tools without per-feature negotiation. Who it does not:a business past eight office seats that needs enterprise-grade access control — the per-user add-on cost climbs fast beyond MAX's included seats.

3. Weave — the most publicly documented ownership event, and a healthcare-first product extending into the trades

What it does: a customer communications platform — voice, text, email and marketing — built primarily for dental, optometry, medical and veterinary practices, serving over 40,000 customer locations, with AI-driven engagement features and a dedicated HVAC industry page extending the product into home services.
What is verifiable:the clearest ownership event in this roster. Weave (NYSE: WEAV) agreed to be acquired by private equity firm Francisco Partners for approximately $650 million, at $7.40 per share — roughly a 34% premium — with the board's approval and a close expected in Q4 2026, subject to shareholder and regulatory approval. Its Q1 2026 revenue reached $65.5 million, up 17.4% year over year, with the company citing AI adoption as a growth driver; Q2 2026 revenue was $67.5 million, up 15.5%.
What is not disclosed:pricing, and no compliance trust page was verified this session. Critically, the product's own customer base concentration is in healthcare, not the trades — buyers should treat the HVAC page as a genuine but secondary market for the company, not its core design center.
Who it fits: an HVAC, plumbing or electrical business that wants a communications platform with a demonstrated AI investment thesis and is comfortable being a secondary market for the vendor. Who it does not:anyone who needs a trades-native roadmap and can't tolerate a pending change of control mid-contract.

4. Jobber — published pricing, and one number we could not reconcile

What it does:field-service management software — scheduling, quoting, invoicing, CRM — with an add-on AI Receptionist that answers inbound calls and captures job details when the office can't pick up.
What is verifiable: published tiered pricing, with solo plans starting near $29/month (Core) and team pricing scaling to roughly $529/month at the top Plus tier (15 users), which includes the Receptionist add-on at no extra cost.
What is not disclosed, honestly: the standalone Receptionist add-on price. Different sources we checked cite $29/month for 30 conversations with a $0.79 per-conversation overage, and others cite a flat $99/month; we could not resolve this to one current, confirmed figure this session. Rather than pick one, we flag the discrepancy — verify the exact number directly with Jobber before budgeting against it.
Who it fits: solo operators and small teams who want transparent, published pricing and are willing to verify one line item directly before signing. Who it does not: a buyer who needs every price point locked down before a first sales call — this is the one vendor in the published-pricing group with a genuine open question.

5. Goodcall — the fastest published setup, with a caller-volume ceiling to watch

What it does:a standalone AI voice agent that answers inbound calls for local service businesses, configured directly from a business's Google Business Profile — location, hours and website — for a fast initial setup.
What is verifiable:published pricing across three tiers — Starter at $79/month (100 unique customers), Growth at $129/month (250), and Scale at $249/month (500) — each bundling unlimited minutes, with a $0.50 per-customer overage beyond the plan's cap. It publishes a documented integration with FieldEdge (vendor #6 below).
What is not disclosed:compliance documentation was not verified this session, and the per-agent unique-customer cap is a real cost variable for a seasonal trade — a heat wave or cold snap can push a business past its tier's cap quickly.
Who it fits: a single-location HVAC, plumbing or electrical business that wants a narrow, fast-to-deploy phone agent without a full platform switch. Who it does not:a business with sharp seasonal call-volume spikes that hasn't modeled the overage cost in advance.

6. FieldEdge (by Xplor) — infrastructure, not an agent, and honest about it

What it does: a 45-year-old field-service-management platform built specifically for HVAC, plumbing, electrical and appliance repair — scheduling, dispatch, invoicing, its own Coolfront flat-rate pricebook, and a two-way live sync with QuickBooks Desktop.
What is verifiable:no native AI dispatcher, AI receptionist, or autonomous booking capability, as of the sources we checked — and to its credit, the vendor doesn't claim one. It does integrate with third-party AI tools, including Goodcall, to add an AI phone layer on top of its dispatch board. Xplor Technologies, its parent, is privately held and backed by Advent International, Battery Ventures and Silver Lake.
What is not disclosed: pricing, and no compliance trust page was verified this session.
Who it fits: an established HVAC or plumbing shop that runs on QuickBooks Desktop and wants its dispatch and pricebook infrastructure decoupled from whichever AI phone layer it chooses — pairing it with #5 above is a documented, real path. Who it does not:anyone shopping specifically for an AI agent — that isn't what this product is, and we rank it honestly as infrastructure rather than pretend otherwise.

7. CallRail — AI Conversation Intelligence, now gated behind a higher tier

What it does: call tracking and marketing-attribution software with an AI Conversation Intelligence layer that analyzes recorded calls for keywords, sentiment and outcomes.
What is verifiable: four published annual-billing tiers — Lead Tracking at $50/month, Lead Tracking Complete at $95, Lead Conversion at $150, and Lead Conversion Complete at $195 — each including 5 local numbers, 250 minutes and 25 texts, with clearly published overage rates. As of the pricing we checked, the AI Conversation Intelligence features require the $150 Lead Conversion tier or above; they are not available as a standalone add-on to the two cheaper tiers.
What is not disclosed:compliance documentation was not verified this session, and the vendor's own site does not enumerate a specific list of home-services FSM integrations — broader integration claims we found came from third-party review sites, not CallRail's own pages.
Who it fits: a marketing-conscious contractor who wants call attribution first and AI analysis as a deliberate upgrade, not a bundled default. Who it does not: a business that only wants the AI layer without paying for the full tracking platform underneath it.

8. Invoca — enterprise-grade, and priced accordingly out of view

What it does: conversation intelligence and call tracking built for larger, multi-location operations across several verticals, including home services, with AI-driven analysis of call outcomes and marketing attribution.
What is verifiable: founding date (2008), headquarters (West Hollywood, California), roughly 390 employees, and a venture capital investor list — Accel, Upfront Ventures, Silver Lake Waterman, H.I.G. Growth Partners and Salesforce Ventures — behind approximately $184 million raised.
What is not disclosed: pricing, which is fully gated behind a sales process; a 30-day free trial is offered instead. No named home-services integrations or compliance certifications were verified on the pages we checked.
Who it fits: a multi-location or franchise-scale trades operation with an existing enterprise procurement process. Who it does not: a single-location shop that will find the sales cycle disproportionate to its size.

9. Broadly — now inside Vendasta, and a real acquisition to track

What it does: a reputation-management, messaging and payments hub aimed at local service SMBs, with AI features that time review requests right after a completed job and speed up responses to inbound web chat.
What is verifiable:the acquisition itself. Vendasta announced its acquisition of Broadly on 17 March 2023, folding Broadly's customer-engagement and reputation-management capabilities into Vendasta's broader SMB marketplace. Vendasta is itself privately held, headquartered in Saskatoon, Saskatchewan. Reviewer data we found shows Broadly's customer base skews heavily toward construction and home services.
What is not disclosed: pricing and compliance documentation, neither verified this session.
Who it fits: a home-services SMB whose main pain point is inconsistent review generation and slow web-chat response, not phone answering or dispatch. Who it does not:anyone evaluating it as a phone-agent or dispatch replacement — that isn't its core function.

10. Rilla — not a phone agent at all, and the vendor most tied to the Cooling-Off Rule

What it does: records, transcribes and analyzes in-personsales conversations — the pitch a technician or sales rep gives standing in a homeowner's kitchen, most often for big-ticket replacement or installation work — and coaches reps against the recording. It does not answer phones or book jobs.
What is verifiable:real, recent capital and partnership activity. A Series B round raised $68.5 million in primary capital, with participants including Bessemer Venture Partners, HubSpot Ventures, PROOF, QuantumLight Capital and Remarkable Ventures. A Home Depot partnership was announced in January 2026, and a Shore Consulting partnership embedding a named sales methodology into Rilla's coaching layer began in July 2025. Third-party research estimates (not vendor-confirmed) put its 2025 private-market valuation between $737 million and $758 million, with a customer base that crossed 2,000 accounts in mid-2025.
What is not disclosed: published pricing; the implied ~$20,000-per-year average contract value we found is a third-party research estimate, not a vendor-stated figure.
Who it fits: a replacement- or installation-heavy business (HVAC system swaps, roofing, home improvement) that runs in-home sales visits and wants to coach and audit those conversations. Who it does not: anyone looking for a phone-answering or booking product — wrong category entirely, and the conversation this tool is built to capture is exactly the one that triggers Cooling-Off Rule obligations discussed later in this article.

11. Podium — the pricing we could not confirm firsthand

What it does: a messaging-first “AI Employee” that responds to leads, requests reviews and processes payments, marketed broadly across local service businesses including the trades; per third-party reviews, its AI capabilities lean toward text and messaging rather than live voice calls.
What is verifiable: corporate history — private, last confirmed at a $3 billion valuation in November 2021, with roughly $425 million raised across investors including Accel, GV, Summit Partners, Y Combinator, Sapphire Ventures, IVP, Album VC and Kickstart Seed Fund.
What is not disclosed, honestly:current pricing. Multiple third-party review sites report Core at $399/month, Pro at $599, Enterprise at $999-plus, and a $99/month AI receptionist add-on, but we could not independently confirm these figures against Podium's own pricing page this session, so we present them as reported rather than confirmed.
Who it fits: a multi-location trades business already comparing several communications and reputation tools where Podium's breadth offsets its price. Who it does not:a single-location shop whose core problem is unanswered phone calls — several reviewers we found specifically note Podium is priced for scale that a single-truck operator doesn't need.

The Binding Constraint: In-Home Sales and the Three-Day Right to Cancel

The rule most home-services owners underrate is not an AI-specific law at all — it's the FTC's Cooling-Off Rule, and it predates every vendor in this roster.Codified at 16 CFR Part 429, it makes it an unfair and deceptive practice for a seller to complete a sale worth more than $25 at a location other than its permanent place of business — a customer's home, most commonly — without giving the buyer a completed, duplicate written cancellation notice and an oral statement of the right to cancel within three business days.

Question	What the rule requires	Who it binds	Practical control
Was the agreement signed at the customer's home, worksite, or anywhere other than your permanent place of business?	16 CFR Part 429 applies to sales over $25 completed at a location other than the seller's permanent place of business	The seller (the contracting business), not the software vendor whose AI priced or pitched the job	Treat every in-home replacement, installation or big-ticket repair agreement as in scope by default
Did the buyer receive both an oral disclosure and a completed written cancellation notice at signing?	The rule requires a completed, duplicate “NOTICE OF CANCELLATION” form in the specified type size, plus an oral statement of the right to cancel	Whoever closes the sale — technician, sales rep, or the business itself if it relies on an app-generated contract	Build the cancellation notice into the signing flow itself so it cannot be skipped under time pressure
Did an AI tool (like a sales-coaching product) price, pitch or draft the agreement?	Irrelevant to the rule's application — the obligation runs to the transaction, not to which tool assisted it	The business closing the sale	Do not treat AI assistance as a reason the disclosure requirement might not apply
Is the recorded sales conversation itself subject to a separate consent requirement?	Independent of the Cooling-Off Rule — state wiretap and recording-consent law governs whether the conversation can be recorded at all	Whoever operates the recording tool (in this roster, most directly Rilla)	Confirm one- vs two-party consent rules for every state you sell in before recording in-home conversations
Within how many days must a cancellation be honored, and what must be returned?	Three business days from the transaction date; the seller must return payments, trade-ins and negotiable instruments within ten business days of receiving the cancellation notice, and release any related security interest	The seller	Build the refund and instrument-return process into your CRM's cancellation workflow with a hard deadline alert

Why this belongs in an article about AI agents: a growing share of the products in this roster — from ServiceTitan's Dispatch Pro routing a technician-turned-sales-rep to a home, to Rilla recording and coaching exactly that in-home pitch — sit directly upstream of a transaction the Cooling-Off Rule governs. None of that changes the rule's application. The obligation runs to the business closing the sale, not to whichever software helped price, pitch or paper it.A contractor who lets an AI-assisted quoting flow move faster than the compliance workflow around it is not reducing risk — they're moving faster toward the same disclosure obligation with less friction to catch a missed step.

Practical takeaway.If your business closes any in-home agreement over $25 — a system replacement, a re-roof, a major electrical panel upgrade — build the Cooling-Off Rule's cancellation notice into the signing flow itself, whether that flow is paper, an app, or something an AI tool helped generate. Treat it as a hard gate the transaction cannot clear without, the same way a payment can't clear without a valid card number.

Bot Disclosure and the Outbound-Call Consent Stack

Two separate legal regimes govern how an AI agent is allowed to talk to a homeowner, and they attach at different moments in the interaction.The first governs whether the agent has to admit it's a bot. The second governs whether it was allowed to call or text in the first place.

California's Bot Disclosure Law, enacted as SB 1001 and codified at Business and Professions Code § 17941, has been operative since 1 July 2019. It makes it unlawful to use a bot to communicate with a California resident online with intent to mislead them about its artificial identity, for the purpose of incentivizing a purchase or sale — unless the operator gives a clear, conspicuous disclosure that it's a bot. A stronger amendment, AB 410, would require the disclosure at first contact and require the bot to answer truthfully if directly asked whether it's human; as of the most recent legislative record we could verify, it was engrossed and held under submission in committee as of 29 August 2025. Check its current status before assuming it has been enacted — the operative law today is § 17941 as originally written.

Channel	The exposure that attaches	What it turns on	Practical control
An AI phone or chat agent trying to book a California caller	Bus. & Prof. Code § 17941 — liability if the bot's operator intends to mislead about its artificial identity for the purpose of incentivizing a sale	Whether a clear, conspicuous bot disclosure is given	Configure a state-aware disclosure that fires for California callers regardless of your primary operating state
An AI agent calling back a missed-call lead	TCPA — FCC 24-17 classifies AI-generated voices as “artificial” voices requiring prior express consent	Whether the call is a service confirmation (lower consent bar) or telemarketing (prior express written consent required)	Classify every outbound call template by purpose before it ships, not after a demand letter
A text or SMS follow-up sent by an AI agent after a missed call	TCPA texting rules run in parallel to the voice-call rules under the same statute	Whether the recipient consented to text communications, and whether the message is transactional or marketing	Capture consent to text at the same point you capture consent to call, not as an afterthought
An in-person sales conversation recorded for coaching (e.g., Rilla)	State wiretap/recording-consent law, independent of the Cooling-Off Rule discussed above	One-party vs. two-party consent by state	Disclose the recording verbally at the start of every visit in two-party-consent states; keep a state-configuration map current as reps travel
Any channel, any state, where the vendor's contract allows it to train its own models on your customers' call or chat data	Contract law, not a specific statute — but it is the highest-leverage clause in any of these agreements	The exact wording of the vendor's data-use and model-training clause	Redline it before signature; require a written data-deletion schedule

Layer the TCPA on top. FCC Declaratory Ruling 24-17 confirms that AI-generated voices fall within the TCPA's restrictions on “artificial or prerecorded voice” calls, meaning an AI agent calling back a missed-call lead needs the called party's prior express consent, and prior express written consent if the call is telemarketing rather than a service confirmation. The commonly misreported wrinkle: Insurance Marketing Coalition v. FCC vacated only the one-to-one lead-generator consent provision of a related 2023 order — not the core written-consent regime for AI-generated calls, which stands. Anyone telling you the TCPA got easier for AI callers is describing a different, narrower ruling than the one that actually applies.

Travel-Time Wage-Hour Rules and Licensing Exposure

Two constraints in this section have nothing to do with AI and everything to do with whether an AI dispatch or scheduling agent is safe to deploy without a human checking its work.

The first is travel time. Under the Fair Labor Standards Act, the Department of Labor draws a clear line: ordinary home-to-first-job and last-job-to-home travel is unpaid commuting, but any travel between job sites during the workday is compensable hours worked — the Department's own guidance uses the example of a plumber traveling customer to customer. If a technician has to stop at the shop first for parts or instructions, that leg is compensable too. A routing or dispatch agent optimizing for fewer total miles or tighter arrival windows can restructure how much of a technician's day counts as paid travel without anyone scoping that as a payroll decision. Treat every routing change a dispatch agent proposes as something payroll needs to see before it ships, not after a wage complaint arrives.

The second is licensing. Electrical, plumbing and HVAC work each carry state-specific (and sometimes municipal) licensing regimes, and the consequences of contracting without the required license are real, if state-specific. California's Contractors State License Board, offered here purely as one state's illustration and not a national figure, treats unlicensed contracting as a misdemeanor carrying up to six months in jail and a $5,000 fine, plus a separate administrative fine ranging from $200 to $15,000. An AI dispatch agent that assigns jobs purely on technician availability, without enforcing license-class matching in its assignment logic, can create exactly this exposure at machine speed — sending an unlicensed or wrongly licensed tech to a job the law reserves for a specific credential.

The pattern to watch for.Both of these are legal questions wearing a software costume. A routing optimization is a wage-hour decision. A dispatch assignment is a licensing decision. Neither should be left entirely to an agent's default logic without a human owning the compliance layer on top of it — and neither is solved by picking a better AI model, because the constraint lives in state law, not in the software.

A Worked Example: 22 Technicians, Three Trades, Two Counties

Consider a 22-technician company running HVAC, plumbing and electrical service out of two locations, with 10 office and dispatch seats needing software access.Rather than print a vendor's unsourced “you're losing $X a year” figure, here is what a real software stack costs, built entirely from the published prices verified above.

Housecall Pro's MAX tier covers up to 8 users at $299/month on annual billing; this company needs 10, so two additional seats at $35/month each add $70, for a base FSM cost of $369/month. Layer Goodcall's Growth tier for after-hours and overflow phone answering at $129/month, covering 250 unique callers; during a summer heat-wave peak the company runs closer to 400 unique callers in a month, adding 150 over the cap at $0.50 each — $204/month in that peak month. Add CallRail's Lead Conversion tier at $150/month to get AI Conversation Intelligence on top of call tracking across 8 marketing-tracked phone numbers (3 over the included 5, at $3 each) and roughly 1,000 minutes over the included 250 (at $0.05 each) — $209/month.

That's $782 in a peak month, or roughly $9,384 a year, for a real, three-vendor software stack — not a vendor's estimate of what you're “losing,” but an owner's own arithmetic from published prices. Compare that number to your own last-90-days call log, the way we recommend throughout this article, rather than to any vendor's calculator.

Now add the regulatory layer. If this same company also sells HVAC system replacements averaging $8,500 during in-home visits — a routine part of the business — every signed agreement over $25 closed in a customer's home triggers the Cooling-Off Rule's three-day cancellation disclosure, discussed in full in our guide to modernizing legacy operational systems around new agent workflows for context on how these compliance layers interact with system integration work. And if this company adopts a tool like Rilla to record and coach those in-home pitches, it now also needs a state-by-state recording-consent map, because two-party-consent states require the homeowner's consent to the recording independent of anything to do with the cancellation right.

What Breaks First

Every failure mode below has a real detection signal and a real rollback — the point is to instrument for it before it happens, not to discover it from a customer complaint.

Failure mode	How you find out	Detection signal to instrument	Rollback
Pricebook drift — the flat-rate book (e.g., a Coolfront-style pricebook) goes stale against real supply costs	Techs quote jobs that lose money, or customers are quoted a price the business can no longer honor	Nightly diff of the agent's quoted-price snapshot against actual invoiced cost per line item	Freeze the agent to read-only pricing until the diff is clean; route quoting to a human in the meantime
License-roster drift — a technician's license lapses or their class changes and the dispatch agent doesn't know	A job is completed by someone whose license status wouldn't have permitted it	Scheduled roster check against state license-board records, or at minimum a quarterly manual audit	Pull the technician from the licensed-work dispatch pool immediately until renewal is confirmed
Seasonal caller-volume spike blows through a per-agent pricing cap	A heat wave or cold snap drives call volume past a plan's unique-caller cap and costs spike or calls start failing over	Monitor unique-caller counts against the plan cap daily during known seasonal peaks	Pre-negotiate a temporary tier upgrade, or route overflow calls to a human queue before the cap is hit
Consent and bot-disclosure configuration drift across states	A new service area opens with the default configuration and no California-specific bot disclosure	Site- or service-area-opening checklist item with a named compliance owner	Disable the voice or chat agent for the affected area until the disclosure configuration is verified
Vendor model change you didn't schedule	Call or chat behavior shifts overnight with no release note	A golden-set regression suite of recorded interactions, replayed on a schedule	Pin prior behavior if the contract allows it; otherwise escalate and add human coverage until the suite passes
Prompt injection through untrusted text (a review, an inbound text, a customer email)	The agent takes an action a normal customer interaction would never trigger	Log every tool call an agent makes and alert on any write action initiated within a session that read external content	Revoke the write credential for that agent identity; injection is unsolved, so the control is blast radius, not detection
Integration disappears in an acquisition or roadmap change	A partner integration quietly stops working after a vendor is acquired (as happened structurally in this category with Broadly's 2023 Vendasta acquisition)	Watch every vendor's changelog and ownership status on a recurring calendar, not just at signing	Keep an exportable call-log and customer-data path so switching cost stays bounded

On that sixth row: prompt injection through untrusted text — a customer text, an inbound email, a review — is not a solved problem for any agent architecture, including every product in this roster. Our full treatment of the issue, including the OWASP LLM Top 10 framing several vendors reference without naming, is in our guide to prompt injection and the OWASP LLM Top 10. The short version for a dispatch or phone agent: treat the control as blast-radius reduction, not detection, because detection alone will miss cases you haven't seen yet.

The Human-in-the-Loop Boundary

The single clearest split in this table is between a bounded, pre-priced job and one that needs an in-home assessment. An agent that reads back a flat rate from a documented pricebook is on safe ground. An agent that commits to a system-replacement price before anyone has looked at the equipment is making a representation nobody has verified — and if that number becomes a signed in-home contract, the Cooling-Off Rule attaches regardless of who or what generated it.

Action	Who may do it	Why the line sits here	Control that makes it safe
Answer hours, service area, and a documented flat-rate diagnostic fee	Agent alone	Retrieval from a source you control, with no open-ended price commitment	Single source of truth for the pricebook; log the record version behind every answer
Book a standard diagnostic or maintenance visit	Agent alone, with a confirmation text or email	Reversible, low-consequence, and the customer has a correction path	Confirmation on every booking; nightly diff against the dispatch board
Quote a bounded flat-rate repair from a published pricebook	Agent alone if the job scope matches a defined line item exactly	This is the technology's strongest use case in the category — a fixed, pre-priced job with no in-home assessment required	Escalate automatically the moment the described job doesn't match a clean pricebook line item
Quote a full system replacement or installation	Never the agent alone — a licensed human must assess and confirm	The number usually depends on an in-home assessment, and a signed agreement over $25 in the home triggers Cooling-Off Rule disclosure obligations regardless of who generated the price	Hard-coded escalation to a licensed estimator for any replacement or installation inquiry
Dispatch a technician to a job requiring a specific license class	Agent proposes, dispatch logic enforces the license match	Assigning unlicensed or wrongly licensed labor to a job is a real regulatory and liability exposure, state by state	License-class data attached to every technician record; the assignment engine hard-blocks a mismatch rather than warning about it
Answer “is it safe to keep running this furnace / is this a gas leak”	Never the agent	A safety-critical judgment call with real physical-harm consequences if wrong	Hard-coded escalation phrase list that routes immediately to a live human or emergency dispatch instructions
Build or adjust a technician's daily route	Agent drafts, an ops manager with wage-hour awareness approves	Between-job travel during the workday is compensable under the FLSA — a routing change is a payroll question wearing a scheduling costume	Log the compensable-travel estimate the route implies and route it past payroll oversight before it ships
Send a follow-up marketing text or make an outbound quote-check call	Agent drafts, consent gating decides whether and how it sends	TCPA consent requirements differ by call purpose and are not the agent's to assume	Consent capture at the point of collection, retained as proof, honored in the same system that sends the message
State that a technician is licensed and insured in the customer's state	Never the agent without a live, verified roster behind it	A false licensure claim is a separate consumer-protection exposure on top of any unlicensed-work penalty	Tie the claim to a roster that is checked against license status on a real schedule, not asserted once and left stale
Publish a response to a negative online review	Agent drafts, a human sends	One bad automated reply outlives every good one, and review text is untrusted input an injected instruction could hide inside	Draft state only inside your own system; no send credential in any session that reads external review text

Two rows deserve special emphasis for this vertical specifically. License-class matching in dispatch is not a nice-to-have configuration option — it is the control that keeps an assignment engine from creating the exact unlicensed-contracting exposure discussed earlier in this article. And route-building is a payroll decision in a trades costume: any change to a technician's stop sequence changes how much of their day is compensable travel under the FLSA, which means an ops manager with wage-hour awareness needs to see it before it ships, not after.

Cost and Timeline

These are Frenchy Digital's standard bands for custom agent work, not a quote for any vendor in this roster's off-the-shelf product.If a published-price product already covers your workflow — Goodcall's $79 starter tier for a single-location phone agent, say — buying it is usually the right call, and we'll tell you that directly rather than propose a custom build you don't need.

Engagement	Range	Timeline	What it covers in a home-services context
Discovery + workflow audit	$9k–$22k	2–4 weeks	Call-volume and booking baseline from your own phone and FSM data, a state-by-state consent and bot-disclosure map, a Cooling-Off Rule process review for in-home sales, and a vendor shortlist with the RFP questions we'd put in writing
Single-workflow agent	$28k–$70k	4–9 weeks	One workflow end to end — inbound call answering, or dispatch assignment, or after-hours booking — with escalation paths, license-match enforcement, and an instrumented review queue
Multi-workflow platform with system integration	$70k–$180k	9–16 weeks	Several workflows across your FSM, call tracking and dispatch board, pricebook drift detection, a golden-set regression suite, state-by-state consent configuration, and an owner-facing reporting pack
Enterprise / multi-location / regulated build	$180k–$420k+	14–24 weeks	Multi-state or multi-brand rollout, per-location isolation, full audit logging with license-match attribution, franchise-level access control, disaster recovery testing, and a documentation package your counsel can review

Every engagement carries a 30-day post-launch warranty and full source-code and IP transfer at completion. Senior-led work runs $150–$225 per hour, ongoing retainers run $2,500–$9,500 per month, and we return a fixed-price phased proposal within 5 business days of a discovery call. For a deeper walkthrough of when a custom build actually beats an off-the-shelf product like the ones ranked above, see our build-vs-buy guide for AI agents.

Red Flags When Evaluating a Vendor

Nine signals worth walking away from, or at minimum pausing on

The vendor won't answer, in writing, what share of calls or bookings its AI completes with zero human involvement
The pricing page runs a “missed call revenue calculator” or an equivalent unsourced dollar-loss figure
A specific accuracy, booking-rate or automation percentage is quoted with no sample size, date range or methodology attached
No named integration with your existing FSM, CRM or call-tracking software — only a generic “integrates with your tools” claim
No stated data retention schedule or DPA, and no willingness to put a deletion timeline in writing
The contract auto-renews with no clearly visible cancellation path in the agreement itself
Nobody on the sales call can explain how the product's dispatch logic enforces license-class matching
No documented handling of the Cooling-Off Rule's three-day cancellation window for any in-home sales workflow the product touches
The sales rep can't describe how the product's bot-disclosure behavior changes for California callers specifically

None of these is disqualifying on its own — plenty of legitimate vendors gate pricing behind a sales call, for instance. But two or more together, especially the human-in-the-loop question paired with an unsourced performance claim, is the exact pattern the FTC's HomeAdvisor order describes after the fact. Ask before you sign, not after.

Limitations and What We Could Not Verify

What we explicitly could not confirm this session

Compliance certifications (SOC 2, PCI DSS or equivalent) for any of the eleven vendors — we found no reachable trust-centre page for any of them and do not assert a certification exists
Podium's current published pricing directly from its own site — the figures in this article are as reported by third-party review sites, not confirmed firsthand
Jobber's exact AI Receptionist add-on price — sources conflict between $29/month and $99/month, and we did not resolve the discrepancy
Rilla's implied average contract value (~$20k/year) and 2025 valuation ($737–758M) — both are third-party research estimates, not vendor-confirmed figures
ServiceTitan's own pricing at any tier — fully gated, with only third-party reseller estimates in circulation that we do not repeat as fact
The current status of California AB 410 beyond the most recent legislative record we could check (held under submission as of 29 August 2025) — verify directly before relying on it
Whether any vendor in this roster holds a specific completed acquisition or funding round more recent than what is cited and dated above

We also want to be explicit about scope: this article ranks phone, dispatch, sales-coaching and communications products serving HVAC, plumbing and electrical contractors specifically. It does not cover general-purpose FSM software without a named AI feature, nor does it cover consumer-facing marketplaces like Angi or Thumbtack, which are a lead-generation category distinct from the AI agents ranked here — though the FTC's HomeAdvisor order, discussed at length above, concerns exactly that adjacent category and is why we treat every vendor claim in this one with the same skepticism.

Want an Honest Read on Your Home-Services AI Shortlist?

Book a free 60-minute discovery call. You leave with a call-volume baseline from your own data, a state-by-state consent map for the areas you serve, and a fixed-price phased proposal within 5 business days.

📅 Schedule Consultation 📞 WhatsApp

1517 S Bentley Ave Unit 204, Los Angeles CA 90025

Top 10 AI Agents forHome Service Contractors in 2026

Table of Contents

Key Takeaways

The Claim Under Test

How We Ranked, and What We Refused to Rank On

What we scored — every attribute re-checkable by the reader

What a Regulator Found the One Time It Audited This Industry's Claims

The Comparison Table

The Eleven, in Order

1. ServiceTitan — the most publicly disclosed company in the roster, on everything except price

2. Housecall Pro — the clearest published pricing and the broadest named AI suite

3. Weave — the most publicly documented ownership event, and a healthcare-first product extending into the trades

4. Jobber — published pricing, and one number we could not reconcile

5. Goodcall — the fastest published setup, with a caller-volume ceiling to watch

6. FieldEdge (by Xplor) — infrastructure, not an agent, and honest about it

7. CallRail — AI Conversation Intelligence, now gated behind a higher tier

8. Invoca — enterprise-grade, and priced accordingly out of view

9. Broadly — now inside Vendasta, and a real acquisition to track

10. Rilla — not a phone agent at all, and the vendor most tied to the Cooling-Off Rule

11. Podium — the pricing we could not confirm firsthand

The Binding Constraint: In-Home Sales and the Three-Day Right to Cancel

Bot Disclosure and the Outbound-Call Consent Stack

Travel-Time Wage-Hour Rules and Licensing Exposure

A Worked Example: 22 Technicians, Three Trades, Two Counties

What Breaks First

The Human-in-the-Loop Boundary

Cost and Timeline

Red Flags When Evaluating a Vendor

Nine signals worth walking away from, or at minimum pausing on

Limitations and What We Could Not Verify

What we explicitly could not confirm this session

Want an Honest Read on Your Home-Services AI Shortlist?

Frequently Asked Questions

Sources & References

Related Articles You May Find Helpful

Chris Machetto

Top 10 AI Agents forHome Service Contractors in 2026

Table of Contents

Key Takeaways

The Claim Under Test

How We Ranked, and What We Refused to Rank On

What we scored — every attribute re-checkable by the reader

What a Regulator Found the One Time It Audited This Industry's Claims

The Comparison Table

The Eleven, in Order

1. ServiceTitan — the most publicly disclosed company in the roster, on everything except price

2. Housecall Pro — the clearest published pricing and the broadest named AI suite

3. Weave — the most publicly documented ownership event, and a healthcare-first product extending into the trades

4. Jobber — published pricing, and one number we could not reconcile

5. Goodcall — the fastest published setup, with a caller-volume ceiling to watch

6. FieldEdge (by Xplor) — infrastructure, not an agent, and honest about it

7. CallRail — AI Conversation Intelligence, now gated behind a higher tier

8. Invoca — enterprise-grade, and priced accordingly out of view

9. Broadly — now inside Vendasta, and a real acquisition to track

10. Rilla — not a phone agent at all, and the vendor most tied to the Cooling-Off Rule

11. Podium — the pricing we could not confirm firsthand

The Binding Constraint: In-Home Sales and the Three-Day Right to Cancel

Bot Disclosure and the Outbound-Call Consent Stack

Travel-Time Wage-Hour Rules and Licensing Exposure

A Worked Example: 22 Technicians, Three Trades, Two Counties

What Breaks First

The Human-in-the-Loop Boundary

Cost and Timeline

Red Flags When Evaluating a Vendor

Nine signals worth walking away from, or at minimum pausing on

Limitations and What We Could Not Verify

What we explicitly could not confirm this session

Want an Honest Read on Your Home-Services AI Shortlist?

Frequently Asked Questions

What is the single best-evidenced reason to distrust a home-services AI vendor's performance claims?

Is there an independent benchmark of AI phone or dispatch agents for HVAC, plumbing or electrical companies?

Does the FTC's Cooling-Off Rule apply to a $9,000 HVAC replacement quote signed in someone's kitchen?

Does California require an AI phone agent to disclose that it's a bot before it tries to book a job?

Do TCPA rules apply to an AI voice agent that calls back a homeowner who missed a technician's arrival window?

Is travel time between job sites during the workday paid, or does it count as an unpaid commute?

Can an AI dispatch agent legally assign any available technician to any job on the board?

Which home-services AI vendors in this roster publish real, dollar-amount pricing?

Do any of these vendors disclose what share of calls or bookings their AI completes with no human involved?

What happened to Weave, and why does its status matter for a home-services buyer?

Is Rilla a phone-answering or booking agent like the others in this roster?

Should an AI agent ever quote a firm price for a system replacement or installation over the phone?

What is the honest cost and timeline for a home-services agent build?

Are “missed call revenue” calculators that vendors post on their pricing pages trustworthy?

What should go in a home-services company's RFP to an AI phone or dispatch vendor?

Sources & References

Related Articles You May Find Helpful

Chris Machetto