AI Receptionist for Service Businesses: Plumbers, HVAC, and Contractors in Chicago

A 3-truck HVAC shop on the northwest side misses an average of two calls a day. The owner is on a roof, the lead tech is in a crawlspace, and the part-time office help left at 3pm. Average ticket is $450. That's $900/day, roughly $4,500/week, or about $230,000/year of work that called the shop and got voicemail. Half those callers will dial the next number on the search results page within ninety seconds. The other half will leave a message and wait — and most won't get a callback fast enough to keep the job. This is the universal Chicago service-business problem: the work is there, the trucks are out, and the phone is the bottleneck.

An AI receptionist solves a specific slice of this. Not all of it — you still need techs who show up, quote fairly, and do clean work — but the slice it owns is the difference between a missed call and a booked job. This post is for plumbers, HVAC, electricians, general contractors, roofers, and handyman services running anywhere from a one-truck operation to a 10-tech shop. Mostly Chicago and the collar counties, but the playbook works in any metro.

The two calls that matter most

Service businesses get a hundred kinds of calls, but two drive most of the revenue and most of the lost revenue:

1. The emergency. "My basement is flooding, it's 2am, I need someone now." These callers are not price-shopping. They will pay your emergency rate. They will book the first qualified company that picks up. If you don't answer, you lose them in under two minutes.
2. The estimate. "I'm thinking about replacing my furnace / redoing my kitchen / rewiring my second floor." These are 15-60 minute conversations. Most don't close. The owner who takes them personally burns half their week on calls that never become jobs.

Different problems, different solutions. AI handles the first by triaging fast and warm-transferring to whoever is on call. It handles the second by qualifying hard up front, booking the in-home estimate (or sending it to a junior estimator) only if the lead is real, and weeding out the tire-kickers without burning the owner's afternoon.

What an AI receptionist actually does on a service-business call

A working setup runs roughly this script for an inbound call:

1. Greet by company name, ask "is this an emergency or are you looking to schedule something?"
2. If emergency: ask one or two qualifying questions (water actively flooding, no heat below freezing, no power, sparking outlet, gas smell), confirm service area, capture name + address + callback, then warm-transfer to the on-call tech with a one-touch approval. If the tech doesn't pick up, escalate to backup. If nobody picks up, text the owner with the call recording and key facts.
3. If schedule: confirm service area (zip check), category of work (drain, water heater, AC tune-up, panel upgrade, etc.), urgency (today, this week, flexible), and book directly to the dispatch calendar.
4. If estimate: capture project scope, square footage or unit type, timeline, decision-maker on the call, and either book a site visit or send to a sales rep with structured notes. SMS the homeowner a confirmation immediately.
5. If outside service area or scope: politely decline and end the call. Don't waste anyone's time.

The whole thing runs in 60-180 seconds for routine calls, and under 90 seconds for emergencies because the AI is built to triage fast and not over-talk.

Service area, urgency, and the "is this real" filter

The single biggest lift an AI receptionist gives a service business is consistent qualification. Humans get tired and start booking everything. AI doesn't.

A reasonable qualification stack for a Chicago HVAC shop:

• Zip code or address check against your service area before anything else
• Equipment age and brand for repair vs. replacement signal
• Decision-maker present for estimate calls (owner-operators won't waste a truck-roll on a tenant who can't approve work)
• Timeline — emergency, this week, this month, "just getting prices"
• Budget signal — not a hard number, but rough scope (one-room mini-split vs. whole-house replacement)

A lead that fails the zip check or the decision-maker check shouldn't book a slot. It should get a polite "we don't service that area" or "let's set up a callback when the homeowner is available." Done right, this alone recovers 5-10 hours a week of dispatcher and owner time.

Dispatch integration: ServiceTitan, Housecall Pro, Jobber

A receptionist that doesn't write to your dispatch system is a glorified answering service. The integrations that matter for service businesses:

• ServiceTitan — the enterprise standard for HVAC, plumbing, electrical. Full API, deep customer/job model. Pricey but every tech and dispatcher knows it.
• Housecall Pro — strong mid-market, especially for owner-operators and 2-10 truck shops. Clean API, good mobile experience.
• Jobber — popular with smaller crews, contractors, and trades that don't need ServiceTitan's depth. Solid API.
• Workiz — common for locksmiths, garage door, appliance repair.
• FieldEdge / ServiceFusion / Service Fusion — niche but real footprint.

The questions to ask any AI vendor:

1. Do you write directly to the dispatch system or just send a CSV/email?
2. Do you respect appointment-type rules (job duration, required skills, two-person jobs)?
3. Do you check tech availability in real time before booking?
4. Do you handle reschedules and cancellations back into the system?
5. What happens when the API is down — do calls drop or does the AI take a message?

A vendor who can't answer those crisply will create more work for your dispatcher than they save.

The emergency call: what good looks like at 2am

Here's the realistic flow for a flooded basement at 2am for a 3-truck plumbing company:

• Caller dials, AI picks up on ring two
• "Plumbing emergency or scheduling?" — caller says emergency
• AI confirms address is in service area, captures name and callback
• AI asks: "Is water actively flowing, or has it stopped?" and "Can you get to your main shutoff?"
• AI calls the on-call tech's cell with a one-button approve
• Tech approves, AI bridges the call, hands off context: "John in Logan Square, 1247 N Whatever, basement actively flooding, main is off"
• If the on-call tech doesn't approve in 30 seconds, AI rolls to backup
• If nobody picks up, AI texts the owner with the recording and pages a third contact

Total elapsed time: under two minutes from the caller's first ring to a tech on the phone. That's the difference between booking the job and losing it. We covered the broader missed-call economics in why missed calls quietly destroy small-business revenue.

A note on warm transfers to cell phones: only enable them for staff who have explicitly agreed and only for the call types they've signed off on. The default for service businesses should be email or SMS dispatch with the tech opting in to live transfers for emergencies they're rostered for. This is the same pattern the trade associations like ACCA and PHCC recommend in their member-facing operations guides.

The estimate call: filtering before the truck rolls

Estimate calls are where margin gets murdered. A kitchen remodel quote eats 30-60 minutes of phone time, then a 90-minute site visit, then a 2-hour proposal write-up. If the close rate is 20%, every booked job carries the cost of four lost ones.

AI doesn't fix the close rate, but it does cut the wasted time. The qualification questions on an estimate call:

• Project type and rough scope (square footage, unit count, room count)
• Timeline (this month, this quarter, "researching")
• Decision-maker on the call or available
• Budget range (not a precise number — a "are you thinking $5k or $50k" signal)
• How they found you (so you know which marketing channel is producing real leads)

A lead that passes all five gets booked for an in-home estimate. A lead that fails on timeline or decision-maker gets a structured nurture sequence — SMS at 7 days, 30 days, 90 days — instead of a truck roll.

SMS follow-up: the cheapest revenue you're not collecting

Most service businesses send a quote and then never follow up. The data on this is consistent across trades: the second touch closes more deals than the first quote. The third closes more than the second. By the fifth follow-up, most shops have stopped — but the leads who close at touch 5 are real revenue.

A reasonable SMS cadence after a quote goes out:

• Day 1: confirmation with quote attached
• Day 3: "Any questions on the quote?"
• Day 7: "Still planning to move forward this month?"
• Day 21: "Wanted to check in — pricing is good for another two weeks"
• Day 60: "Are you ready to schedule, or should we close this out?"

AI runs this on autopilot. Replies route to a human. It's the single highest-ROI automation a service business can add. We dug into the underlying math — including why the first 60 seconds matter so much — in our breakdown of lead response time and AI.

Cost: AI vs. answering service vs. another hire

A typical month for a 2-10 truck Chicago service business looks like this:

AI isn't a replacement for a great office manager. It's a replacement for the answering service plus the after-hours gap plus the lunch-hour coverage gap. Most shops we work with run the AI alongside their existing front-office staff — humans take daytime calls, AI takes overflow and after-hours.

What to avoid

A few patterns that look good in a demo and fall apart in production:

• AI that talks too much. Service callers want speed. If the AI runs a 90-second greeting before asking a question, callers hang up.
• No human fallback. Every AI setup needs a "press 0" or "let me talk to someone" path that actually works.
• No service-area gate. Booking jobs in zip codes you don't serve creates dispatcher chaos.
• No recording or transcripts. If you can't audit what the AI told the caller, you can't improve it and you can't defend it if a customer complains.
• No fallback when the dispatch system is down. AI should take a clean message if the API is unreachable, not drop the call.

Bottom line

A 3-truck shop missing two calls a day at $450 a ticket is leaving roughly $230,000 a year on the table — an AI receptionist that costs $300-$700 a month and books straight to dispatch closes most of that gap. The setup that works is fast triage, hard qualification, real CRM integration, and a clean handoff to a human the moment the call needs one.