For Lehigh HVAC I built a custom internal CRM (customer management, job tracking, follow-ups, campaign lists, and call outcomes) as the system of record for inbound and outbound ops, plus Retell AI for front-desk calls (scheduling, FAQs, intake, urgent escalations) and outbound duct-cleaning campaigns that replaced manual office dialing.
HVAC companies do not lose work because they lack technicians. They lose work because the phone is busy, the front desk is understaffed, and new services never get a consistent outbound pitch.
For Lehigh HVAC, I built a home-services automation stack in three connected parts: an internal CRM as the system of record, Retell AI for the front desk, and Retell outbound for a newly launched duct cleaning offer.
1. Internal CRM - system of record
The CRM was not a side database for the voice agents. It was the operating system for both inbound and outbound work.
What I built:
| Module | Purpose |
|---|---|
| Customer management | Single profile per household/business the office and agents can reference |
| Job / service tracking | Active installs, service calls, and status through the lifecycle |
| Follow-up management | Callbacks and open items that do not live in someone’s notebook |
| Campaign lists | Segmented lists for marketing outreach (including duct cleaning) |
| Call outcomes | Transcripts, disposition, and next steps written back from Retell agents |
Because inbound front-desk calls and outbound duct-cleaning campaigns both read from and write to the same CRM, the office could report on volume, interest, and booked jobs without reconciling spreadsheets after every shift.
Voice AI without a CRM is a novelty. Voice AI wired into CRM is an operations upgrade.
2. Retell AI front-desk agent
The next layer was inbound coverage - replacing repetitive front-desk phone work so staff could focus on installs, dispatch exceptions, and complex service.
The Retell agent handled:
- Incoming customer calls during peak AC/heat season and after hours
- Appointment scheduling and callback booking
- FAQ responses (hours, service area, common HVAC questions)
- Capturing customer information (name, address, issue type, urgency)
- Escalating urgent HVAC issues to a human when the situation is an emergency or outside the script
Why Retell for this client profile: HVAC owners and office managers need to edit the agent after launch - hours, greeting, transfer numbers, offer language - without opening an engineering ticket for every copy change. Retell is the stack many home-service operators prefer for that reason.
Under the hood: strong prompts, HVAC vocabulary, confirmation loops for addresses and appointment windows, and explicit handoff rules when a human must take over.
3. Duct cleaning outbound campaign
When Lehigh launched duct cleaning, the bottleneck was not the service - it was manual outreach. Office staff did not have hours to dial existing customers between front-desk rings.
Retell outbound agents:
- Called existing customers from CRM campaign lists
- Pitched the new duct-cleaning service with a consistent script
- Handled common objections (price, timing, “not right now”)
- Captured interested leads and next steps
- Pushed results back into the CRM so humans could book the job
That replaced manual calling by office staff and shortened the path from “new service launch” to “booked duct cleaning jobs.”
How the three parts connect
Inbound call ──► Retell front desk ──► CRM (customer + job + outcome)
Outbound list ──► Retell campaign ──► CRM (lead + disposition)
Office staff ──► CRM dashboard ──► Book jobs, handle escalationsMake.com (and similar workflow glue) syncs events between Retell and the CRM where native hooks are not enough - so a hang-up still lands as a record even when the caller never spoke to a human.
Results shape (what “good” looks like)
For Lehigh-style HVAC + CRM + voice builds, success shows up as:
- Fewer missed front-desk calls during peak season and after hours
- Shorter time from new service launch to booked jobs
- Office staff on exceptions, not every ring
- One CRM that inbound agents, outbound campaigns, and humans all trust
Earlier HVAC voice work in this portfolio also measured ~20% higher satisfaction and shorter support calls when conversational design was dialed in - same discipline applies here: domain prompts, natural TTS, and context retention on the phone.
Stack notes
| Layer | Choice |
|---|---|
| Client | Lehigh HVAC |
| Operations | Custom internal CRM (system of record) |
| Voice | Retell AI inbound front desk + outbound campaigns |
| Conversation | GPT-style prompting + tuned TTS where needed |
| Orchestration | Make.com - outcomes sync to CRM |
| Offer motion | Duct cleaning outbound to existing customer lists |
Who this is for
If you run HVAC, roofing, plumbing, or a similar home-service company and you are tired of:
- Missed calls that become Google reviews for someone else
- Launching a service with zero systematic follow-up
- Paying for a front desk that cannot scale on peak days
- Voice demos that never write to your CRM
…this is the playbook: CRM as system of record + Retell inbound + Retell outbound.
Related work
Common questions
What did you build for Lehigh HVAC?
A custom internal CRM (customers, jobs, follow-ups, campaign lists, call outcomes) plus Retell AI for inbound front-desk coverage and outbound duct-cleaning outreach - the CRM was the system of record for both directions.
What did the internal CRM include?
Customer management, job and service tracking, follow-up management, campaign lists for marketing outreach, and storage of call outcomes from the Retell agents so inbound and outbound shared one source of truth.
What did the Retell front-desk agent handle?
Incoming customer calls, appointment scheduling, FAQ responses, capturing customer information, and escalating urgent HVAC issues to humans - especially to cut missed calls during busy seasons and after hours.
How did the duct cleaning outbound campaign work?
When Lehigh launched duct cleaning, Retell outbound agents called existing customers, pitched the new service, handled common objections, captured interested leads, and pushed results back into the CRM - replacing manual dialing by office staff.
Do you build CRM and voice agents for other home-service companies?
Yes. The same pattern - custom or GHL CRM plus Retell inbound/outbound - fits HVAC, roofing, plumbing, and similar trades that live and die by phone speed-to-lead and front-desk coverage.