Public AI Voice Case Studies in HVAC Dispatch: From 43% Missed to 97% Answered

Avoca's AireServ + Wilson Companies cases, ServiceTitan's HVAC AI rollout, and the published booking-rate lifts that took small dispatch shops from 55% to 90% in weeks.

The customer / use case

HVAC, plumbing and electrical contractors lose money the moment a call rings out. Industry data: roughly 6 in 10 calls to small home-service businesses go unanswered, and ~80% of those callers will not leave a voicemail — they call the next company on the SERP. AI voice agents target that lost-call gap directly: cover after-hours, cover overflow, route emergencies, book the job in the contractor's FSM (ServiceTitan, Housecall Pro, FieldEdge).

flowchart LR
  C[Call] --> AV[Voice agent]
  AV --> EM{Emergency?}
  EM -->|Yes| ON[On-call tech SMS + bridge]
  EM -->|No| BK[Capture address + symptoms]
  BK --> ST[ServiceTitan job created]
  ST --> CFM[SMS confirmation + tech ETA]
  CFM --> AN[Post-call review + win-back tag]

What they did

• Avoca AI / AireServ and Wilson Companies ran AI on after-hours overflow. After-hours booking rates jumped from 40–55% (human answering services) to 85–95% with AI.
• One contractor profiled by Whistler Billboards moved from 43% missed calls to 97% answered in the first week.
• Another HVAC company recovered 14 booked jobs in week one — jobs that would have rolled to voicemail with the old answering service.
• A service-business profiled by ServiceTitan switched from a legacy answering service and pushed booking rate from 55% to 90%, which forced them to hire more techs to cover the dispatch board.
• Vendor-aggregated metrics: 15× revenue uplift, 3× higher conversions, 80% CSAT lift, with payback usually inside the first 10 captured jobs (a 20-truck shop spends $1,500–$3,000/mo on the agent vs. ~$300+ recovered per booked job).

Outcomes (real numbers)

• 43% missed → 97% answered (one HVAC operator, week 1).
• 55% → 90% booking rate post-switch (ServiceTitan-published case).
• 14 jobs recovered in week 1 of an HVAC rollout.
• After-hours booking 85–95% with AI vs 40–55% with human answering services.
• $1,500–$3,000/mo cost typically paid back inside 10 jobs.

CallSphere comparable build

Explore a live demo and compare current plans to find the right fit for your business.

Behind the agent are CallSphere's standard 37 agents · 90+ tools · 115+ Postgres tables — the FSM job creation tool, the on-call tech tool, the address-validation tool, the upsell tool (filter, plan, membership), and the recall-list tool that fires automated maintenance-season callbacks.

FAQ

Will my dispatchers lose their jobs? No. Every HVAC case study above ended up hiring more techs because the booked board grew. Dispatchers move to higher-margin work: tech routing, recall list, customer save calls.

Can the agent triage real emergencies (gas leak, no heat)? Yes — the emergency-triage tool checks for known emergency keywords, escalates to the on-call tech, and bridges the call. CallSphere ships this preconfigured with the contractor's escalation protocol.

What if my FSM isn't on the supported list? We support ServiceTitan, Housecall Pro, FieldEdge, Jobber, ServiceFusion natively. For others (mHelpDesk, Service Autopilot) we ship a generic webhook + Zapier path on the Pro tier.

Pricing math: when does it pay back? At $499/mo + 10 booked jobs/mo at $300 average ticket = $3,000 captured vs $499 cost = 6x return on cost. Real Avoca/Wilson numbers are higher because emergency tickets average $600+.

Sources

• ServiceTitan — "AI Voice Agents in HVAC: The Contractor's Complete Guide (2026)" — https://www.servicetitan.com/blog/ai-voice-agents-in-hvac
• ServiceTitan — "Understanding the HVAC AI Landscape in 2026" — https://www.servicetitan.com/blog/hvac-ai
• Avoca AI — homepage + case stories — https://www.avoca.ai/
• Whistler Billboards — "AI Assistants Are Answering Service Calls" — https://www.whistlerbillboards.com/marketing/ai-assistants-are-answering-service-calls/
• Famulor — "AI Voice Agent for HVAC and Trades: 24/7 Job Dispatch" — https://www.famulor.io/blog/ai-voice-agent-for-hvac-and-trades-247-job-dispatch

How this plays out in production

To make the framing in Public AI Voice Case Studies in HVAC Dispatch: From 43% Missed to 97% Answered operational, the trade-off you cannot defer is channel routing between voice and chat — a missed call should not die, it should warm up the SMS or web-chat lane within seconds. Treat this as a voice-first system from the first prompt: the agent's persona, its tool surface, and its escalation rules all flow from that single decision. Teams that ship fast tend to instrument the loop end-to-end before they tune any single component, because the bottleneck is rarely where intuition puts it.

Voice agent architecture, end to end

A production-grade voice stack at CallSphere stitches Twilio Programmable Voice (PSTN ingress, TwiML, bidirectional Media Streams) to a realtime reasoning layer — typically OpenAI Realtime or ElevenLabs Conversational AI — with sub-second response as a hard SLO. Anything north of one second of perceived silence and callers either repeat themselves or hang up; that single number drives the whole architecture. Server-side VAD with proper barge-in support is non-negotiable, otherwise the agent talks over the caller and the conversation collapses. Streaming TTS with phoneme-aligned interruption keeps the cadence natural even when the user changes their mind mid-sentence. Post-call, every transcript is run through a structured pipeline: sentiment, intent classification, lead score, escalation flag, and a normalized slot extraction (name, callback number, reason, urgency). For healthcare workloads, the BAA-covered storage path, audit logs, encryption-at-rest, and PHI-safe transcript redaction are wired in from day one, not bolted on at compliance review. The end state is a system where every call produces a row of structured data, not just a recording.

FAQ

What does this mean for a voice agent the way Public AI Voice Case Studies in HVAC Dispatch: From 43% Missed to 97% Answered describes?

Treat the architecture in this post as a starting point and instrument it before you tune it. The metrics that matter most early on are end-to-end latency (target < 1s for voice, < 3s for chat), barge-in correctness, tool-call success rate, and post-conversation lead score distribution. Optimize whatever the data flags as the bottleneck, not whatever feels slowest in your head.

Why does this matter for voice agent deployments at scale?

The two failure modes that bite hardest are silent context loss across multi-turn handoffs and tool calls that succeed in dev but get rate-limited in production. Both are solvable with a proper agent backplane that pins state to a session ID, retries with backoff, and writes every tool invocation to an audit log you can replay.

How does the After-Hours Escalation product make sure no urgent call is dropped?

It runs 7 agents on a Primary → Secondary → 6-fallback ladder with a 120-second ACK timeout per leg. If the primary on-call does not acknowledge inside the window, the next contact is paged automatically — voice, SMS, and push — until somebody owns the incident.

See it live

Book a 30-minute working session at calendly.com/sagar-callsphere/callsphere-llc-meeting and bring a real call flow — we will walk it through the live after-hours escalation product at escalation.callsphere.tech and show you exactly where the production wiring sits.