It is 103 out and you are flat on your back in an attic with a flashlight in your teeth and insulation down your collar. The phone buzzes in your pocket. You are not answering it. Twenty minutes later you climb down, drink half a bottle of water, and there are four missed calls and one voicemail that says nothing but "hey, give me a call back." Two of those people already booked somebody else. You will never know which two.
That is the actual problem. Not "call handling." Not "customer experience." Just money walking out the door on the hottest week of the year while you are doing the work. So you start searching, and Goodcall comes up fast. It is one of the more visible self-serve AI phone agents out there, it has a real pricing page you can read without booking a call, and it has a Google origin story attached to it. Reasonable place to start looking.
Fair warning before you read further: this is written by Thermoi, and Thermoi sells a competing service. So the deal here is that we are not going to pretend Goodcall is bad. For plenty of businesses it is a sensible pick, and for some HVAC shops it will be the right call over us. What we will do is lay out what Goodcall publishes about itself, explain why an HVAC phone line behaves differently from a salon phone line, and let you decide which side of that line your shop sits on.
What Goodcall is, and who it actually fits
Goodcall sells a self-serve AI voice agent that picks up your inbound calls, answers common questions, captures lead info, books appointments, and hands the call off to a human when it hits something it cannot finish. They label it "Agentic Voice AI" and say you can have a custom phone agent live in a few minutes with no engineering team involved. Their About page tells the origin story: the idea started inside Google Ads in 2017, came out of a Google hackathon, launched as CallJoy under Google Area 120, then spun out into its own company. Their listed backers include Neo, Foothill Ventures, Merus Capital, Xoogler Ventures, Spencer Rascoff, and AI people from Microsoft, Amazon and Google. No dollar amounts are published.
Setup is a five step flow on their site: connect your business, define the agent's skills, set conversation logic, choose your phone setup, go live. The vocabulary is worth knowing before you sign up. "Skills" are the things the agent can answer. "Flows" split into forms, which collect lead and qualifying info, and logic, which branches based on a yes or no, a multiple choice answer, or whether the caller is new or returning. "Team members" get dashboard access. "Directory contacts" receive transfers and notifications but cannot log in. Out of the box the agent is your first line and picks up everything, but you can run it as overflow instead by setting a human answer window, meaning it only takes the call after your line rings a set number of times. There is also a customizable operator fallback: by default it transfers to your business line, and you can change that to a specific person, a department, taking a message, sending a self-service link, or scheduling a callback.
On phone setup, every Goodcall agent comes with its own assigned number in a local area code you choose, and if your existing number matters to your brand, their documented path is conditional call forwarding from your current carrier. They say the agent also works on an existing Google Voice number once you verify it. Per their help center, service is currently available in the USA only, with an interest form for people outside it.
The pricing is public, which is genuinely useful and more than a lot of this category offers. Billed monthly, per agent: Starter is $79, Growth is $129 and is marked most popular, Scale is $249. Annual billing knocks 15% off and brings those to $66, $108, and $208 a month. All three tiers include unlimited minutes and tokens, which sounds enormous until you look at what actually meters. Goodcall bills on unique customers, which their FAQ defines as any caller with a unique phone number who calls and interacts with your agent in a given month. Starter allows 100 a month, Growth 250, Scale 500, and after that it is $0.50 per additional unique customer. Robo calls, numbers you block, and callers who never say anything do not count against the allowance. The tiers also differ on logic flows (1, 3, or 25), team members (3, 9, or 50), directory contacts (3, 25, or 500), and how long call and customer detail is retained (7 days, 30 days, or unlimited). There is a fourth Enterprise tier that is quote only, and that is where the dedicated account manager, custom API integrations, SLA and priority support, and custom onboarding live. Nothing about SLAs or onboarding help is promised on the self-serve tiers.
One thing to hold onto if you are price shopping: that billing model is unusual. Most services in this space charge you per minute or per call. Goodcall charges per unique caller per month, per agent. Comparing $129 against somebody's per-minute rate on a spreadsheet will mislead you in both directions depending on your call mix, so figure out roughly how many distinct phone numbers dial your shop in a month before you decide anything is cheap or expensive.
So who is Goodcall genuinely right for? If you run a one truck or two truck operation, your call volume is modest, most of your calls are routine, and you would rather sit down on a Tuesday night and configure the thing yourself than sit through a demo, Goodcall fits that person well. If you want a price on a page before you talk to a human being, it fits. If you run more than one location or more than one business, they say there is no limit to how many agents you can build, so one agent per location is on the table. And their claimed compliance posture, which they list on their homepage as SOC 2 Type II, ISO 27001 and HIPAA, is more than most small vendors put in writing. Take that as their claim rather than something we verified, since their trust center did not render for us. Same goes for their headline numbers. Goodcall says 42,000 businesses, 50,000 agents launched, and more than 60 million voice interactions. Those are self-reported marketing figures with no date stamp or methodology attached.
Here is the caveat that matters most for you specifically. Goodcall is deliberately horizontal, and they say so themselves: businesses of all sizes across every industry. Look at where their own site actually points and you see salons, restaurants, BPO and contact centers, plus named case studies in fitness, junk removal, beauty, and cookies. The vertical hubs in their own footer are salons and BPO. HVAC and plumbing do show up on their site, but in templated SEO pages and blog posts, not in their navigation, product pages, or pricing. That is not a knock on the software. It just means nobody built the thing around a July heat wave and an on-call rotation.
One more practical note. The integration path Goodcall documents on its own product and pricing pages is Zapier, which they describe as natively embedded and which they say opens up 10,000 or more tools. Their Zapier app itself exposes three triggers (customer lookup, flow complete, call complete) and one action. Microsoft Teams, Genesys and HubSpot are named on their Enterprise page, in the context of custom deployments, and custom API integrations are a line item on the quote-only tier. If you need a booking to land inside a specific field service platform, do not assume it does. Ask them directly, get the answer in writing, and apply that same standard to us.
Why an HVAC phone line is a different animal
A salon books a haircut. Worst case, the appointment gets moved. Nobody's pipes freeze. The reason generic answering tools struggle in the trades is not voice quality or booking mechanics, it is that a large share of your calls require a judgment call in the first thirty seconds, and getting that judgment wrong costs real money in one direction or real goodwill in the other.
Start with triage. A caller saying the upstairs is a little warm and a caller saying there is no cooling at all in a 100 degree house with a ninety year old mother in the back bedroom are the same words to a call taker following a script and completely different jobs to you. A tenant reporting a burning smell is not the same as a homeowner asking about a maintenance plan. Water dripping through the ceiling below the air handler needs somebody today, or the drywall bill gets added to the repair bill. Somebody who says they smell gas should not be booked at all, they should be told to get out and call the gas company. That is not a nice-to-have branch in a flow chart. That is the whole job of picking up the phone at your company.
Then there is when the calls come. Furnaces fail on the coldest night, not at 10am on a Wednesday. Heat pumps quit during the heat wave. A meaningful chunk of your best jobs, the ones people pay premium rates for without arguing, come in outside business hours, and that is exactly when a person is least likely to be sitting at your desk. If your after-hours plan is a voicemail box, you already know how that goes. People do not leave messages anymore. They hang up and dial the next shop on the map pack.
Peak season breaks things differently again. The first genuinely hot week of the year does not increase your call volume gradually, it stacks a week's worth of calls into two days while your techs are already booked out. The bottleneck is not that calls go unanswered in a general sense, it is that they all arrive at once, and whoever normally answers is now also dispatching, ordering parts, and fielding a supplier who wants payment. Something gets dropped. It is usually the call from the person who was ready to buy.
And then there is the on-call rotation, which a horizontal tool leaves you to assemble yourself out of logic branches, transfer rules and contact lists. You do not want your on-call tech woken at 1am for a filter question or a price shopper. You absolutely want him woken for a no-heat call in January at a house with a newborn in it. The value of after-hours answering is not that something answers, it is that it correctly decides which of those two calls is worth a phone ringing on a nightstand. Get that wrong in the permissive direction and your best tech quits. Get it wrong in the restrictive direction and you lose the emergency ticket that pays for the week.
There is a booking layer on top of all of it. The information you actually need before a truck rolls is not a name and a phone number. It is the address, whether it is residential or commercial, owner or tenant, what the system is and roughly how old, whether it is under warranty, whether they are on a maintenance agreement, and whether anyone will be home. A tool that takes a message and emails it to you has moved the work, not done it. The booking has to close while the person is still on the phone, because the moment they hang up they are calling the next shop.
Where Thermoi fits
Thermoi answers inbound calls for HVAC contractors, 24 hours a day. It picks up, qualifies the call, books the appointment while the caller is still on the phone, and routes genuine emergencies to whoever is on call. Every call produces a full transcript, so you can read exactly what was said instead of guessing what your answering service told a customer at 2am. That is the whole product. It was built for one trade, which is the only real argument we are making here.
The entry point is a $50 pilot for two weeks, running on your actual phone line with your actual customers. Not a sandbox, not a demo number. If it does not earn its place in two weeks on real calls during your real week, you are out fifty bucks and you learned something. Monthly plans start at $299. We do not publish the tiers above that, and we should be straight about why that is a tradeoff and not a virtue: the plan gets sized on a 15 minute demo based on your call volume, because a shop taking 60 calls a month and a shop taking 900 are not the same customer. The honest cost of doing it that way is that you cannot compare us on a page at midnight. If that alone is a dealbreaker, Goodcall publishes their numbers and that is a legitimate reason to go with them.
No setup fees. No contract. Month to month, cancel when you want, and a 90 day money-back guarantee behind it. Most contractors are live in under 48 hours.
Now the part that costs us something to say. Thermoi is early and pre-revenue. We have no customer count to quote you, no case studies with a revenue lift number in them, no logos, no awards, and we are not going to invent any. If you want a vendor with years of published references behind it, we are not that yet, and Goodcall has been around since the CallJoy days. You are weighing a service built specifically around the way your phone rings against a company with a longer track record and a public price list. That is a real tradeoff and pretending otherwise would be the first lie in the relationship. The $50 pilot exists precisely because we know we cannot ask you to take that on faith.
How to actually pick between them
Forget features for a second and answer three questions about your own shop.
First: what share of your calls need a judgment call rather than an answer? If you mostly get people asking about hours, pricing, and scheduling a routine tune-up, a horizontal tool configured well will handle that, and Goodcall is a strong candidate. If a real slice of your calls are people in distress at bad hours and you need something to sort the genuine emergency from the person who is annoyed but fine, you want something built around that sorting, and you should be looking at trade specific options.
Second: who is going to build and maintain it? Goodcall's model is that you do it. Skills, flows, logic branches, the operator fallback, the human answer window, all of it is yours to configure and yours to fix when a caller says something you did not anticipate. Some owners genuinely like that and will have it dialed in by the weekend. Others will set it up in a hurry in June, never touch it again, and quietly stop trusting it by August. Be honest about which one you are. If it is the second one, do not buy the self-serve product and blame the software later.
Third: how much does after-hours actually matter to your revenue? Pull your call logs and count what came in between 6pm and 7am over the last two months, then estimate what you closed from that window. If the number is small, this is a convenience purchase and you should buy on price. If the number is large, this is a revenue purchase, the difference between $79 and $299 stops being the interesting variable, and what matters is whether the thing wakes the right tech for the right call.
A practical way to run it: pick the two you are seriously considering and put each one on your real line for two weeks. Read the transcripts. Not the dashboard, the transcripts. You will learn more from twenty real conversations with your own customers than from any comparison anybody writes, including this one. Goodcall's pricing page metadata advertises a free trial, though the terms are not spelled out in the visible page copy, so ask them what it includes before you count on it. Thermoi's is the $50 two-week pilot. Either way, test on live calls in your own market. Everything else is guessing.
Keep reading
The side by side version of this lives on the Thermoi vs Goodcall comparison. For what an AI answering service actually does on each call, start with the HVAC answering service overview, and the pricing page has the $50 pilot and the ROI calculator on it.
Call patterns differ by market. There are breakdowns for Phoenix, Houston, Dallas, Miami, and Atlanta.
Questions people ask before they switch
Is Goodcall good for HVAC companies?
It can work, especially for a small shop with mostly routine calls, but it is not built for the trade. Goodcall describes itself as serving businesses of all sizes across every industry, and its own navigation, product pages, and case studies point at salons, restaurants, BPO and contact centers, fitness, junk removal and food. HVAC and plumbing appear on their site mainly in templated SEO pages and blog content, not in the product or pricing. If your calls are mostly hours, pricing and routine scheduling, that may not matter. If a real share of your calls need emergency triage and on-call routing, you will be building that logic yourself.
How much does Goodcall cost?
Goodcall publishes three self-serve tiers, priced per agent. Billed monthly it is $79 for Starter, $129 for Growth, and $249 for Scale. Annual billing is advertised at 15% off, which works out to $66, $108 and $208 per month. All three include unlimited minutes and tokens, but they cap unique customers at 100, 250 or 500 per month with $0.50 per unique customer after that, so your real cost tracks how many distinct callers dial you, not how long you talk. There is a fourth Enterprise tier that is quote only with no published price.
Does Goodcall integrate with ServiceTitan, Housecall Pro or Jobber?
Do not assume so. The only integration path Goodcall documents on its own product, pricing and how-it-works pages is Zapier, which they describe as natively embedded and which they say connects to 10,000 or more tools, plus support for Google Voice numbers. The Goodcall app in Zapier's own directory lists three triggers and one action. Their site does have pages naming various field service platforms, but the language on them is conditional rather than a statement that the integration exists today. If a booking has to land inside your dispatch board automatically, ask them to confirm it in writing before you pay. Ask any vendor the same thing, including us.
What is the cheapest way to test AI call answering on my real phone line?
Test on live calls, not a demo number, and keep the commitment short. Thermoi's entry point is a $50 pilot that runs for two weeks on your actual line, and there is no contract or setup fee behind it. Goodcall's pricing page metadata advertises a free trial, though the terms are not stated in the visible page copy, so ask them what it covers. Whichever you try, read the call transcripts afterwards rather than the summary dashboard. That is where you find out how it handled the caller who was upset.
Can an AI actually handle a no-heat emergency at 2am?
It can handle the intake and the routing, which is most of what goes wrong at 2am. Thermoi answers, works out whether it is a genuine emergency or something that can wait for the morning, books it, and pushes the real emergencies to whoever is on call, with a full transcript of what was said. What no AI does is decide for you where the line sits between an emergency and a nuisance. That is your policy, and it needs to be written down clearly before you turn anything on, because the whole value of after-hours answering is waking the right person for the right call and leaving your tech alone for the rest.