Pull the call logs for any trades shop running three to thirty trucks and the same number keeps showing up. In one 2026 analysis of 130,175 inbound calls across 45 contractors, 74.1 percent went unanswered. That is three out of four people with a broken furnace, a leaking water heater, or a dead panel, ringing through to voicemail. And 85 percent of callers who hit voicemail never call back. They dial the next name on the list.
This is not a staffing problem you can hire your way out of. The calls cluster at 7am before techs roll, at lunch, after 5pm, and all weekend. Your office manager cannot answer the phone and dispatch and chase parts at the same time. So the question for this month is narrow and answerable. What does it cost to put a voice agent on the line for the hours your people cannot cover, and does it book enough work to pay for itself?
Call answer rate is the metric, not “AI”
Trades owners are right to be skeptical of “AI” as a category. So ignore the category and watch one number. Call answer rate is the share of inbound calls a live voice picks up before voicemail. Most shops think theirs sits around 80 percent. The call-log studies say the real figure during peak season is closer to 50 percent, and worse on weekends.
The dollars are easy to size. Davinci Virtual’s 2026 home-services research pegs the cost of a missed service call at 800 to 1,200 dollars in lost work. The ACHR News analysis put the annual bleed at an average of 189,068 dollars per contractor. You do not need those exact figures to hold. Take your own average ticket, multiply by the calls you miss in a week, and apply your booking rate. A shop missing twelve calls a day at a 300 dollar average and a 35 percent close rate is leaving about 1,260 dollars on the table every day. That math is why voice agents stopped being a novelty.
What an after-hours voice agent actually does
The useful tools are not answering machines with a better voice. A voice agent answers on the first or second ring, identifies the caller’s problem, asks your qualifying questions, quotes your after-hours trip fee, checks your service area by ZIP, and writes a booked appointment straight into your field-service software. The good ones also triage. A caller who says “I smell gas” gets routed to your on-call line right away. A caller who wants a maintenance quote gets booked for Tuesday.
The reason this matters for trades specifically is that the agent takes action, not just messages. A receptionist service that emails you a transcript at 11pm still leaves the booking for the morning, by which point the customer has hired someone else. An agent that writes the job into Housecall Pro or ServiceTitan while the caller is still on the phone captures the work before the next shop ever gets a chance.
Two honest limits. First, these agents handle the standard 85 to 95 percent of calls cleanly, scheduling, pricing, service-area checks, and emergency triage, and pass the weird ones to a human. Second, they are only as good as the script and pricing you feed them. Garbage trip-fee rules in, garbage quotes out. Budget an afternoon to set the rules, not five minutes.
The stack and what it costs
There are two roads. Use the AI built into the field-service platform you already run, or bolt on a dedicated voice vendor. Rough 2026 pricing follows so you can size it against your own missed-call math.
If you run Housecall Pro, its CSR AI answers calls, books jobs, and logs customer details directly in your account with no separate integration to wire up. Housecall Pro prices CSR AI on request rather than publishing a flat rate, so call your rep with your call volume in hand. The base plans for reference run 59 dollars a month for Basic, 149 for Essentials, and 299 for MAX on annual billing.
If you run ServiceTitan, the native path pairs with answering vendors built around its API. AnswerForce is the common pick for higher-volume HVAC shops because it writes back into ServiceTitan and ships HVAC-specific intake scripts rather than generic ones. ServiceTitan’s own Dispatch Pro then assigns the booked job to the right tech by drive time, skill match, and job value, so the after-hours booking lands clean on the morning board.
If you want a flat, predictable bill independent of your CRM, the dedicated voice vendors compete hard on price. Local Call AI runs a flat 297 dollars a month for unlimited calls. Others like Thoughtly and LeadTruffle sit near 199 dollars a month and integrate with the major platforms. At 199 a month, capturing a single 3,500 dollar replacement job pays for more than seventeen months of service. That is the whole investment case in one sentence.
Set it up without wrecking your dispatch
The failure mode is not the technology. It is dropping a robot voice in front of your best customers with no rules. Do it in this order and you avoid that.
First, decide the hours. Most shops start the agent after-hours and weekends only, keeping the daytime human line untouched. That alone recovers the biggest pool of missed calls with the least risk. Second, write the triage tree. List your true emergencies, gas, no heat in a freeze, active leak, sparking panel, and route those to your on-call cell, not to a Tuesday slot. Third, load your real pricing. After-hours trip fee, service-area ZIPs, and the three or four job types you actually book by phone. Fourth, set the handoff. Any call the agent cannot resolve goes to voicemail or your cell with the transcript attached, so nothing vanishes.
Then test it yourself before it ever touches a customer. Call your own number at 9pm as a panicked homeowner with no heat. Call again as a price shopper. Call a third time with an address two counties away to confirm it declines politely. If all three behave, turn it on for one weekend and read every transcript Monday morning. You will find two or three script gaps. Fix them and run another weekend. Inside three weeks the agent handles your after-hours calls better than the rushed voicemail it replaced.
The one decision to make this month
Do not try to pick the perfect vendor. Pick a starting point and measure. If you already run Housecall Pro or ServiceTitan, turn on the native AI answering for after-hours only and read the first month of transcripts. If you are on a smaller CRM or none, start a 199-dollar-a-month dedicated agent on a 30-day trial and point only your weekend calls at it.
Before you flip it on, pull last month’s call report and write down two numbers: total inbound calls and how many your team actually answered. That is your baseline answer rate. Thirty days later, pull the same report. If the agent moved your answer rate from 50 to 90 percent and booked even four extra jobs, it paid for a year of itself and you stop guessing. The shops winning the peak-season rush in 2026 are not the ones with more trucks. They are the ones whose phone gets answered at 7pm on a Sunday.
At Atlas Unchained we build this kind of after-hours intake stack for trades shops that would rather book the call than babysit the tool. If you want help wiring the triage tree to your real pricing and CRM, that is exactly the work we do.
