Disclosure: Futuro sells the outbound AI receptionist described on this page, so we have a commercial interest in its conclusion. No other vendors are named here; this is a capability deep-dive, not a comparison. Capabilities attributed to Futuro are published company claims, date-stamped August 2026. Every dollar figure in the install-base model is an illustrative assumption, labeled where it appears. The compliance section describes legal frameworks generally and is not legal advice.
Most of what is written about AI in the trades is about answering: the missed call, the after-hours emergency, the surge week. This article is about the other direction. Your install base, every homeowner who ever paid you to put in a system, is the most valuable list your company owns, and in almost every shop it sits unworked because nobody has the hours to call seven hundred people. Futuro’s outbound AI receptionist works it: a service-first check-in call, your approved warranty or service-agreement offer, and a payment link by text or email to close. This page explains exactly how that works, what the math looks like with assumptions visible, and where the legal lines run.
Everything below was researched and written in August 2026, last reviewed August 28, 2026, per our editorial standards. Compliance frameworks are cited to the primary sources (the FCC, the FTC, the statute text) rather than to summaries of summaries, and the things we did not do are stated plainly in the methodology.
About this article: This is not a listicle and not a comparison; it explains one Futuro capability (outbound calling) for one industry (HVAC) and one use case (warranty and service-agreement programs). No competing products are mentioned, because the contrast that matters is not against another tool; it is against the status quo, which is that nobody makes these calls at all. Futuro is the only vendor discussed, and we say so plainly. External sources are cited throughout, and the agreement-economics figures carry their years.
How many of your install base have never been asked?
Start with the arithmetic that motivates this entire article, with every assumption labeled so you can swap in your own. Illustrative model: your company installed 900 systems over the last five years. Assume 15% of those homeowners bought the warranty or service agreement at the point of sale. That leaves 765 homeowners who were never asked again: not because they declined, but because nobody called. Assume a $240 average residential agreement and a deliberately cautious 10% yes rate on a well-run outbound pass. That is roughly 76 new agreements, about $18,000 in first-year recurring revenue, from a list you already paid to create. None of those four numbers is a cited statistic; they are dials. Set them to your reality and the conclusion survives almost every setting. The model itself is published as a CC-BY dataset, so you can reuse the arithmetic with attribution.
The highest-margin product you already own
The industry’s own numbers explain why the unworked list matters. Recurring service agreements captured 55% of HVAC service revenue in 2024, a segment growing at about 8.3% a year, per FieldEdge’s 2025 program guide. Preventive maintenance contracts accounted for 39% of total U.S. HVAC services revenue in 2025, per Mordor Intelligence’s market reporting as summarized in PipelineON’s 2026 analysis, and the same analysis puts the average residential HVAC customer’s lifetime value at $15,340, with plan-attached customers worth multiples of that. Contractor Mike Rosone’s framing, featured by BuildOps and quoted in that piece: a dollar of maintenance agreement tends to pull one to three dollars of downstream repair and replacement work. And when owners sell, 2025 M&A aggregation by Main Street Wealth puts companies with recurring contract bases at 5 to 8 times EBITDA against 2 to 4 times for purely transactional shops. The agreement is not a coupon book; it is the valuation.
Why the list never gets worked
Ask any owner why 765 past customers have not been called and you get the same three answers. Techs do not pitch: they are paid to fix systems, not to sell, and the Bureau of Labor Statistics projects roughly 40,100 HVACR openings a year through 2034, so every good hire goes to a truck, not a phone. The office cannot call: your CSR is already doing the work of two people during the season. And there is the owner’s guilt: everyone knows the list is money, and nobody has the hours. A human working 765 names at a realistic daily pace of live conversations (illustrative assumption) needs weeks for one pass; the campaign never survives contact with a heat wave.
The two windows when homeowners say yes
Timing does more work than persuasion in this campaign. The first window is the post-install month: the homeowner just spent five figures with you, trusts you, and is thinking about protecting the purchase; a check-in at day 30 is service, and the agreement offer lands naturally inside it. The second is the pre-season window: September, before the first heating call, when a tune-up and priority-service promise answers a fear the homeowner already has. An outbound pass timed to either window converts on the homeowner’s own logic, which is why the call is framed as a check-in rather than a pitch.
What does the outbound call actually sound like?
Annotated beat by beat, because the design is the entire product. The call below is the configured pattern; your onboarding session sets the specifics.
Beat one: the identification, disclosed by design
The call opens with the truth, per your configured disclosure settings and applicable law: the agent’s name, that it is an AI calling on behalf of your company, and why it is calling. The FCC’s 2024 ruling treats AI voices as artificial voice under the TCPA, so identification is not a courtesy; it is the architecture. Some states require specific disclosure language; the exact wording is set at onboarding with your counsel’s input.
Beat two: the check-in framing
The call is a customer-service touch before it is anything else: how is the system running, any issues we should know about, is everything performing the way it should. This is deliberate design, not etiquette. A homeowner who hears genuine interest in their equipment stays on the line; a homeowner who hears a pitch in sentence one does not. The program offer only makes sense after the check-in, because the check-in is what makes it your company calling rather than a telemarketer.
Beat three: the personalization nobody can fake
“We installed your Trane in March of 2023.” That sentence is the whole advantage, and it is impossible for a cold caller to counterfeit: the agent carries the install record into the call (equipment, date, any warranty terms that shipped with the unit). A stranger offering a warranty is spam. The company that put the system in, calling to check on it by name and model year, is service. Homeowners hear the difference in the first ten seconds.
Beat four: the condition questions
Before any offer, the agent asks the questions your best CSR would: is the system cooling and heating evenly, any unusual noise, how are the utility bills versus last year. The answers do double duty. They feed your service pipeline (a homeowner describing a struggling compressor is a booked diagnostic), and they make the subsequent offer feel like a recommendation instead of a script, because it now is one.
Beat five: the transition to the offer
Only after the check-in does the offer arrive, and it arrives as housekeeping: “a lot of our customers with a system your age protect it with our service agreement; would you like to hear what it covers?” Permission-based, one sentence, and a genuine off-ramp. If the homeowner says no here, the call completes as a pure service touch, which is a win in itself: you just reminded 765 people who installed their system.
Beat six: the offer, exactly as approved
The offer the agent presents is the offer you configured: what the program covers, the price, the terms, nothing else. Futuro’s MasterMind layer bounds the conversation to your approved program documents, so the agent cannot improvise a discount, invent a coverage term, or quietly promise the moon to close. What your office approved is what 765 homeowners hear, word for word, on every call, at 9 AM and 8:45 PM alike.
Beat seven: the questions, answered from your documents
“What does it cover?” “Is the compressor included?” “What does it cost after year one?” The agent answers from the program documents loaded into its knowledge graph, the same source of truth your office uses. If a question falls outside the documents, the honest answer is a scheduled callback from your team, not a guess. This is where the 94% human-indistinguishable voice from our published 1,000-participant study earns its keep: the conversation holds together under real questions, in 53 languages.
Beat eight: three close paths
Every call ends one of three ways, and all three are designed. Yes: the agent sends a payment link by text or email from your own payment platform (Stripe or whatever you already run); the customer completes enrollment on their own phone, and the campaign logs the conversion end to end. No card numbers are ever spoken on the call. Maybe: the agent texts or emails the program PDF and books a specific callback time, which lands on your campaign calendar rather than in a memory. No: logged, thanked, and suppressed per your frequency rules, never argued with. The relationship is worth more than the enrollment.
What are the compliance rules, honestly?
This is the section most outbound tooling never writes, and it is the reason this article exists in this form. Outbound AI calling is legal, and it is regulated. Both things are true, and a shop that understands the framework runs a better campaign than one pretending the framework is optional.
The FCC has already ruled on AI voices
In February 2024 the FCC adopted a declaratory ruling (FCC 24-17) confirming that AI-generated voices are an “artificial or prerecorded voice” under the Telephone Consumer Protection Act, effective immediately; the news release and independent legal analysis lay out what that means. Practically: outbound AI-voice calls need prior express consent (and for marketing content, prior express written consent), identification of who is calling, and a working opt-out. The statute’s private right of action runs $500 per violating call, trebled if willful, with no aggregate cap, which is why the consent records matter as much as the script.
Consent and the Do-Not-Call list
Two habits do most of the work. First, consent capture: collect calling consent at the point of sale and on every form, so the install-base list carries its own permission trail. Second, registry hygiene: telemarketers selling goods or services must scrub against the National Do Not Call Registry, per the FTC. Your install base has an advantage here worth understanding precisely: an existing business relationship changes the telemarketing analysis in many frameworks (Florida’s statute, for example, carves prior or existing business relationships out of “unsolicited” at § 501.059(1)(k)), but it does not remove the TCPA’s AI-voice consent requirement for marketing calls. Translation: your own customers are the right list, and you still configure consent and suppression correctly.
Calling windows and state rules
The federal baseline limits outbound telemarketing calls to 8 AM to 9 PM local time at the called person’s location, per the Telemarketing Sales Rule (16 CFR Part 310), with parallel delivery restrictions in the FCC’s own rule at 47 CFR 64.1200; the FTC’s business guidance summarizes the whole rule set in plain language. States can and do run narrower: Florida’s Telephone Solicitation Act requires prior express written consent for automated sales calls and carries its own $500-per-violation damages with trebling, and other states set their own windows, registries, and disclosure language. The campaign configuration encodes your states’ rules before the first dial.
What Futuro configures and what you own
The division of labor, stated plainly. Futuro configures: the disclosure language at call open, the calling-window rules by area code and state, the opt-out and suppression handling, the frequency caps, and the consent-field mapping on your list. You own: the decision to call, the list itself, the consent language on your paperwork, and the final legal review with your counsel. Nothing on this page is legal advice, and any vendor who tells you compliance is fully “handled” without your counsel is a vendor to walk away from.
How does the campaign run as a channel?
The reframe that matters: this is not a one-time blast, it is a measurable marketing channel that happens to use the phone. Channels have inputs, cadence, a close, and reporting.
The list: upload and segment
The campaign starts from your data: an export of the install base with name, number, install date, and equipment. Segmentation is where the money hides: 2022–2024 installs get the post-install check-in, pre-2020 equipment gets the aging-system framing (the efficiency-upgrade conversation ENERGY STAR’s equipment guidance primes homeowners for), lapsed agreement holders get the win-back. Each segment hears a different reason for the call, because each segment is a different homeowner.
The cadence: persistent, polite, capped
One pass is not a campaign. The configured cadence retries no-answers at different times of day inside the legal windows, leaves the follow-up text where appropriate, and caps total attempts per household so persistence never becomes pressure. Caps are a feature, not a limit: the list is an asset you will call again next season, and households that feel respected answer next time.
The close: a payment link by text or email
The close is the part owners disbelieve until they see it. On a yes, Futuro (our product) sends a payment link by SMS or email from your own payment platform; Stripe Payment Links or whatever you already run. The homeowner taps, pays, and is enrolled; the money lands in your account through your existing merchant relationship. No card numbers spoken aloud, no invoicing lag, no “we’ll mail you something.” For a shop that has never worked its install base, the first campaign week genuinely behaves like found money: revenue from customers you already paid to acquire, closed by infrastructure you already had.
The reporting loop
Every call is logged and the campaign reports like a channel: connects, offers made, enrollments, and revenue, visible in your analytics dashboard and summarized in a digest your office can read in ninety seconds. You will know the yes rate by segment by the end of week one, which means the second campaign is tuned by evidence instead of hope.
When does outbound pay for itself fastest?
The post-install 30-day check-in
The highest-trust moment you will ever have with a homeowner is the month after you installed their system. An automated check-in on every install, running continuously, asks how the system is running and offers the agreement while the purchase is still fresh. This campaign never finishes; it becomes part of how your company installs.
The pre-season agreement push
September is the agreement month: the first heating call of the season is coming, and priority service plus a pre-winter tune-up answers a fear the homeowner already holds. A segmented pass over the unenrolled install base in early September, before your phones detonate, fills the shoulder-season board with maintenance visits and the year with renewals.
The lapsed-agreement win-back
The cheapest enrollment in your market is the homeowner who already bought the agreement once and let it lapse. They understand the product, they liked it enough to buy, and something administrative (a moved autopay, a missed renewal notice) ended it. A polite win-back pass converts at multiples of a cold list for the same reason the check-in works: it is not a stranger calling. (The five-figure lifetime value in PipelineON’s agreement analysis is why the lapse list is worth the dial.)
What can’t outbound fix?
The honest limits, because they exist and you should hear them from us.
A weak program
If the agreement is overpriced, thin on coverage, or worse than a competitor’s, the AI will not sell it at scale; it will simply discover that fact faster and with better logging. Outbound amplifies the offer you have. If the agreement needs tightening, the trade’s own quality benchmarks, like ACCA’s published standards, are the reference point: fix the offer first, and the campaign rewards you immediately after.
A stale list
Disconnected numbers, moved homeowners, and landlines that now belong to someone else all cap the connect rate, and no voice quality fixes bad data. A list export from your FSM is usually clean; a five-year-old spreadsheet may not be. The reporting loop shows you the connect rate in week one, so list problems surface as numbers rather than vibes.
A skipped compliance setup
The frameworks above are not decoration. A shop that dials without consent records, suppression handling, or calling-window rules is not running a growth channel; it is accumulating statutory exposure at $500 a call. Futuro configures the controls and your counsel reviews the posture; both halves happen, or the campaign does not dial. And some customers simply do not want sales-adjacent calls at all, which is what the frequency caps and the permanent no are for: the point of the design is that the program protects the relationship even when the answer is no.
How did we research this page?
We publish this page and sell the product it describes, so the method notes carry the load.
Evidence level: founder-confirmed capabilities, frameworks independently cited
Evidence level for this page: the outbound capabilities described (list-based campaigns, install-memory personalization, offer bounding, payment-link close, campaign reporting) are founder-confirmed, generally-available product capabilities, date-stamped August 2026. The agreement-economics figures come from dated third-party industry sources linked inline, each carrying its year. The compliance section cites the primary frameworks (the FCC ruling, the TCPA statute text, the TSR and FCC rules in the eCFR, the FTC’s registry guidance, and Florida’s statute as a state example) rather than commentary about them. The 765-homeowner model is illustrative arithmetic with every assumption labeled, not a measured result.
What we did not do
What we did not do: we did not run a live customer campaign for this article, and no enrollment or revenue figures here are measured campaign results. We did not record the annotated call; the beat structure is the configured call pattern, and a staged, labeled recording will be added to this page when it exists. We did not provide, and this page is not, legal advice; the compliance section describes frameworks generally and your counsel owns your campaign’s review. And we did not compare competing outbound tools, by design: the comparison that matters is against the status quo, in which nobody makes these calls at all.