Takeaways from the Service Council webinar with John Carroll (Service Council), Denny Lawrence (Comfort Systems USA), and Edwin Pahk (Aquant), with research insights from Gerardo Pelayo.
TL;DR - 7 Big Things You Should Know
- Voice changes trust and speed. Studies cited by Service Council suggest voice interactions can increase trust and lower cognitive load for users on the move, perfect for field service.
- Techs are still hunting for info. In Service Council’s latest Voice of the FSE research, technicians report spending up to 40% of their day searching for information; fewer than half say knowledge bases are easier to use than five years ago.
- Comfort Systems’ “Fix Center” is the keystone. Eight senior experts centralize tribal knowledge and coach techs, already supporting ~13,000 cases and ~28,000 calls YTD 2024, with 36% of techs using it (across ~3,600 total techs).
- Now adding Voice AI. Comfort Systems is piloting Aquant’s Voice AI (their agent is nicknamed “Mark”) to deliver hands-free troubleshooting and better data capture, then escalate cleanly to a human expert when needed.
- Grounded answers only. The voice agent references uploaded manuals and known sources; if it’s unsure, it hands off to an expert technician, reducing hallucination risk and avoiding “do it twice” moments.
- Data fuels accuracy. After expanding from ~20 documents to 700+ manuals (plus basic HVAC primers), Comfort reports internal test accuracy around 60–70% so far, with a plan to keep climbing as usage and data increase.
- Next: smarter call routing. Comfort is designing a voice-first IVR replacement to route customers to the right team, answer status questions automatically, and triage BAS/technical support, doing more with the same staff while improving CX.
Why Voice AI, and Why Now?
Field engineers are always in motion. Gloves on, tools out, phone buried, yet they need answers now. Service Council’s research highlights a persistent gap: knowledge is exploding (new technology, complex install bases, rising expectations), but access to that knowledge hasn’t kept pace. Voice offers a more human, lower-friction interface:
- Natural: Ask questions conversationally rather than drilling through menus or apps.
- Faster: No hunting through PDFs or tapping around with cold fingers.
- Trusted: Intonation and dialogue build confidence, and when the AI doesn’t know, it says so and transfers.
As John Carroll put it, voice can be more inclusive and accessible, aligning with how humans actually work, especially when they’re mobile.
About the company:
Comfort Systems USA is a national MEP (mechanical, electrical, plumbing) contractor with an ~$8B business, including more than $1B from service operations.
Phase 1: The Fix Center — Scaling Human Expertise
Anticipating the “silver tsunami” of retiring experts, Comfort Systems created the Fix Center — a team of eight senior technicians tasked with capturing tribal knowledge and supporting field techs in real time.
- Scale so far: ~13,000 support cases and ~28,000 calls YTD 2024
- Adoption: 36% of ~3,600 field technicians actively use it
- Knowledge base: Troubleshooting guides, triage flows, repeat failure patterns, and parts histories
The Fix Center became Comfort’s single source of truth — both a safety net for newer techs and a training ground for the next generation.
Phase 2: Voice AI with Aquant: Extending the Fix Center’s Reach
To make that expertise instantly accessible in the field, Comfort is piloting Aquant’s Voice AI (nicknamed Mark), a conversational assistant designed for hands-free troubleshooting and data capture.
- Knowledge foundation: Grounded in 700+ OEM manuals plus basic HVAC primers
- Hands-free experience: Techs talk to Mark through a headset while continuing to work
- Smart escalation: When Mark reaches the limit of its knowledge, it transfers the case to a Fix Center expert with a structured summary — no repeated steps or lost context
- Continuous learning: Every interaction is captured and can be converted into new knowledge artifacts, compounding improvement over time
What Improved Accuracy
Early AI tests struggled with complex troubleshooting. The breakthrough came when Comfort added basic HVAC training materials alongside OEM manuals, giving the AI both the language of the field and the language of the manufacturer. That combination enabled the system to connect how technicians describe issues with how documentation defines them, a replicable approach for any service organization.
Phase 3: Voice-First Customer Care — The Next Frontier
Building on its success with internal support, Comfort is now reimagining the customer experience. The goal: replace rigid IVR systems with a voice-understanding front door that can instantly recognize intent and route calls to the right destination.
- Route based on caller type and issue (customer status, vendor question, tech support, BAS)
- Answer simple questions automatically (e.g., work-order status)
- Conduct first-tier troubleshooting using the same knowledge base before escalation
Expected impact:
- Higher CSAT from faster answers and fewer transfers
- More productivity from the same headcount
- Cleaner data and fewer duplicate interactions
Building for Trust: How Comfort Is Rolling Out Change
For Comfort Systems, technology change wasn’t just about deploying AI – it was about earning trust from the technicians who’d rely on it every day. So the rollout began where trust already existed: inside the Fix Center.
Before the first field pilot, senior experts stress-tested the AI with real cases, identifying knowledge gaps and refining responses. Only once the system could reliably handle everyday HVAC scenarios (grounded in 700+ manuals and basic primers) did Comfort invite technicians into the mix.
Early adopters became champions. Newer techs helped veterans adapt to digital workflows, while experienced techs shared their field wisdom to strengthen the model. The result: a rollout that felt less like a software launch and more like a shared evolution of the craft.
This approach directly answers challenges surfaced in Service Council’s research: technicians still spend up to 40% of their day hunting for information, and while efficiency has improved, mental fatigue hasn’t. By designing a system that reduces friction and cognitive load — and builds confidence rather than replacing human judgment — Comfort is closing that gap.
The Bottom Line
Comfort Systems USA isn’t just piloting another AI tool — it’s redefining how knowledge moves through the service organization:
- From manuals to conversations
- From tribal expertise to institutional memory
- From IVR menus to intent-aware routing
With hands-free access, grounded answers, and smooth escalation to experts, voice AI is helping every technician work faster, think clearer, and serve customers better — without burning out the people who make it all possible.
What Other Service Leaders Can Learn
Comfort’s journey offers a simple but powerful roadmap for anyone exploring Voice AI in service:
- Start with one problem, not one tool. Comfort didn’t chase trends — they solved a real issue: techs spending too much time searching for answers.
- Ground the AI in your experts. Before scaling, they built trust internally, letting their most experienced technicians shape and test the system.
- Blend foundational and specialized knowledge. The breakthrough came from combining training materials with OEM manuals — helping the AI understand how people actually talk in the field.
- Design for escalation, not perfection. “I don’t know” isn’t a failure; it’s a feature that keeps the tech honest and humans in the loop and able to provide input when it needs it.
- Measure what matters. Accuracy, time-to-resolution, and first-time fix — but also technician confidence and mental load.
Voice AI isn’t just a shortcut to speed — it’s a bridge between how people work and how technology learns
Thanks to Service Council and Comfort Systems USA for a candid, forward-looking session. The webinar recording can be found here and Aquant’s ebook, “Building an Enterprise‑Ready Voice AI: Your 8‑Step Playbook” can be downloaded here.