Yes. I think there's a real opportunity—but I'd frame it differently.
The opportunity isn't "AI for plumbers." It's building the operating system for field service businesses, where AI happens to be the interface.
The reason this market is interesting is that trades have three characteristics that AI is unusually good at:
- Most valuable information is unstructured (phone calls, texts, photos, voice notes, invoices).
- Every company has slightly different workflows, making traditional software rigid.
- Labor is scarce, so saving an hour of office work has immediate ROI.
Where I'd be cautious is assuming AI alone is the moat. It probably isn't.
The real customer problem
If you shadow an HVAC owner for a week, they aren't saying:
"I wish I had GPT."
They're saying things like:
- "Where did Mike put that estimate?"
- "Did we ever bill this job?"
- "Why is this technician driving across town?"
- "Which maintenance customers are due next month?"
- "How much margin are we making?"
- "Can someone answer the phone?"
Those are operations problems.
AI is valuable because it can remove the human bottlenecks.
I would design it around one idea:
Every interaction becomes a structured job record automatically.
Imagine this flow.
Technician arrives
Instead of opening six apps, they hit one button.
"Started job."
The AI logs:
- arrival time
- GPS
- customer
- truck inventory
- weather
- previous service history
without typing.
During the repair
The technician naturally speaks.
"Bad capacitor. Replaced with a 45/5. Condenser coil dirty. Recommended cleaning. Customer declined."
The AI converts that into:
- service notes
- warranty records
- inventory deduction
- customer explanation
- invoice line items
No paperwork.
Customer asks questions
The technician takes a picture.
AI responds:
"Heat exchanger has visible cracking."
Then generates:
- repair recommendation
- replacement estimate
- financing options
- email summary
Leaving the job
Instead of sitting in the truck for twenty minutes...
Invoice is already drafted.
Customer signs.
Payment collected.
Office instantly knows job is complete.
The office experience
This is where I'd spend most effort.
Imagine opening the dashboard.
Instead of spreadsheets...
The AI says:
"Three invoices haven't been sent."
"Two maintenance contracts expire this month."
"Truck #6 is missing inventory."
"One technician is consistently taking 40% longer on capacitor replacements."
That's much higher leverage than another chatbot.
Suppliers
Supplier negotiation is interesting but I'd do it later.
First:
AI notices:
"You buy capacitors from three vendors."
Then:
"Vendor B is 12% cheaper."
Eventually:
"Should I generate a purchase order?"
Later:
"I negotiated pricing with your preferred distributor."
That requires trust and integrations.
I wouldn't make it a launch feature.
Predictive maintenance
Photos are valuable.
But history is even more valuable.
Imagine combining:
- photos
- equipment age
- climate
- previous repairs
- manufacturer
- technician notes
Then AI predicts:
"78% chance this compressor fails within 12 months."
That's compelling because it helps contractors sell legitimate preventative maintenance rather than reactive repairs.
The biggest mistake I'd avoid
Trying to replace the existing software.
Every HVAC company already has something like:
- ServiceTitan
- Housecall Pro
- Jobber
- FieldEdge
- QuickBooks
Those systems are deeply embedded.
Replacing them means asking owners to migrate years of customer history.
That's painful.
Instead...
Become the AI layer on top.
Phone
↓
Voice
↓
AI
↓
ServiceTitan
QuickBooks
Jobber
Google Calendar
Gmail
Inventory
CRM
This is a much easier sale.
What becomes your moat?
Not the language model.
Everyone has access to similar models.
The moat is proprietary operational data.
Imagine collecting millions of:
- repair photos
- technician voice notes
- equipment failures
- invoice outcomes
- maintenance histories
- supplier pricing
- job durations
Eventually your system learns things no generic model knows.
Example:
"This brand of compressor installed in Texas in 2018 usually fails after X hours."
That's difficult for a competitor to replicate quickly.
Biggest pitfalls
1. Technicians hate extra steps.
Every tap loses adoption.
Voice should always beat typing.
2. Hallucinations are dangerous.
AI cannot invent:
- parts used
- labor hours
- diagnoses
- invoices
Treat AI as a drafter that always lets the technician review critical records.
3. Connectivity is unreliable.
Many job sites have poor reception.
Offline-first operation with background syncing is essential.
4. Trust matters more than intelligence.
Owners don't care that the AI is "smart."
They care that:
- payroll is correct
- invoices are accurate
- taxes reconcile
- jobs don't disappear
Reliability is a product feature.
5. Start with one trade.
HVAC, plumbing, electrical, and landscaping share similarities, but each has different terminology, workflows, compliance needs, and seasonality.
Nail one vertical before expanding.
If I were starting this company today
I'd focus on one promise:
"Your technicians never do paperwork again."
Version 1 would have just four core capabilities:
- Voice-first job capture: Record conversations and convert them into structured service notes, parts, labor, and customer summaries.
- Automatic back-office workflows: Generate invoices, estimates, purchase orders, and accounting entries for review.
- Operational copilot: Surface overdue invoices, scheduling conflicts, inventory shortages, maintenance renewals, and profitability insights proactively.
- Deep integrations: Sync with the systems contractors already use rather than replacing them.
That positioning is easier to explain, delivers measurable ROI through reduced administrative work and faster billing, and creates a foundation for more advanced capabilities like predictive maintenance, supplier optimization, and autonomous scheduling as customers build trust in the platform. The long-term vision isn't just another AI assistant—it's a system that quietly keeps the entire business running while owners focus on serving customers and growing the company.
