On weekends, Angie Snow’s family didn’t buy movie tickets without a plan.
If the business phone rang, she and her husband would have to step into the hallway to take it.
“My husband called that our ball and chain,” Snow told a packed session at AHR Expo 2026 in Las Vegas, drawing knowing nods from HVAC contractors who share the same after-hours grind.
Snow, Principal Industry Advisor at ServiceTitan and a former owner who built her shop through rigorous call center discipline, made the case for AI voice agents — not as a replacement for CSRs, but as a scalable way to stop missing revenue during peak season.
“Humans are limited,” she said. “How can we use these new resources and new tools that are available to help?”
The Call Center Crunch — and the Trust Gap
From April spikes to winter surges, the busy season exposes the same bottlenecks: abandoned calls, CSR hiring and burnout, and after-hours coverage that rarely meets the standard.
“Calls are getting abandoned,” Snow said. “You most likely paid for a lead for that call, and you’ve just lost that money.”
Third-party after-hours services often disappoint, too: “They’re not your people.”
Yet the bigger obstacle is trust.
“Customers hate automation,” Snow acknowledged, citing years of clunky IVRs and dead-end menus. The remedy isn’t to reject automation — it’s to improve it. “Automation is not the problem. We just need better automation,” she said.
Her reframing lands squarely on outcomes, not technology.
“The goal is to capture 100% of our inbound calls,” Snow said. “That is not the goal [for] 100% AI.”
The model: let AI handle volume and filtering, then route nuanced or escalated cases to human CSRs.
Effective AI in the call center must do three things:
Capture the opportunity: identify the caller, address, and issue.
Move the call forward: book the job or trigger a timely human follow-up.
Protect the customer experience: preserve the tone and trust you’ve built.
How to Deploy (and Win): From Pilot to Scale
Snow’s field-tested implementation advice is blunt: “Do not turn everything on all at once.”
Start with one lane — after-hours, overflow, or follow-ups. Limit job types (e.g., maintenance first), and “make sure there is a clear escalation path to a human.”
Setup matters. “You’ve got to have clean job types,” she said. Define capacity by job type, write scripts that reflect your brand, and codify booking rules, fees, and hours. Equally critical is change management.
CSRs will ask, “Am I being replaced by AI? Do I still matter?” Snow’s answer: “It’s to capture every lead… It’s not to replace them. It’s to empower them.”
The operational musts:
Get the basics right:
Clean job types, with clear rules for which jobs AI can book vs. escalate.
Capacity by job type (service, maintenance, sales) defined in advance.
Booking rules (hours, policies, fees).
Scripts that match your brand. “If you’re just putting garbage into the voice agent, that’s what’s gonna come out.”
Lead with transparency:
Start every call by stating the caller is speaking with an AI voice agent and always offer an immediate path to a human. Don’t “dupe” customers; it erodes trust.
Use a three-phase framework:
Pilot: Choose one lane (after-hours, overflow, or follow-ups). Limit job types. Create a clear escalation path to a human.
Prove: Review calls daily, tune the script and tone, and tighten routing with an IVR that sends calls to CSRs first during business hours and to AI on overflow or after-hours.
Scale: Expand to confirmations, reschedules, and more job types. Upload your knowledge base so the AI reflects your policies, pricing, and process.
Measure What Matters—and the Payoff
Snow urged contractors to track four domains:
Capture: Abandoned call rate, response time, percent of calls handled by AI vs. humans.
Conversion: Booking rate, time to book, cancellation rate (watch for changes after deploying AI).
Experience: Escalation reasons and volumes, repeat calls, customer reviews.
Efficiency: CSR load, schedule coverage when staff is out, and office-to-field ratio.
For ServiceTitan users, native voice agents raise the ceiling.
Because the agent lives inside the platform, it “can already read our dispatch board,” Snow said, matching jobs to technician availability and skills for touchless, accurate booking.
The performance data is compelling.
“A 70% booking rate—this is overall,” Snow reported, even when capacity is tight. “When there’s room in the schedule, it’s booking up to 90% of those calls.” One early adopter saw its voice agent book more than 500 jobs in a month — 23% of total bookings.
From a 1,000-contractor survey ServiceTitan ran, Snow highlighted where the trades stand on AI: 12% are already embedding it into workflows, 34% are experimenting, 19% are interested but not using it yet, and 22% are waiting and observing. The common pain points AI can address are familiar—abandoned calls, CSR burnout, after-hours coverage, and inconsistent experiences.
The north star, Snow reminded the room, isn’t replacing people. It’s never missing a call.
“AI can help you handle the volume, move calls forward, and protect the customer experience,” she said. “Use it to make your humans even better.”
