Your next customer might be a robot

CALL RECORD              TUE 7:42 PM
CALLER    assistant acting for homeowner, Champaign IL
REASON    water heater, leaking, tonight if possible

19:42:07  dialing shop 1 of 4 .. no answer, 6 rings
19:42:51  dialing shop 2 of 4 .. voicemail, declined
19:43:12  dialing shop 3 of 4 .. answered in 0.8s
19:43:58  window 8-10am confirmed, deposit authorized
19:44:02  report to homeowner: "Handled."

HUMANS ON THE CALL: 0

That call has not happened yet. Give it two or three years.

Read it again, though, because the most expensive line is not in it. There were four shops on that list. The third one answered, so the fourth was never dialed. That plumber was at dinner with his family. His phone never rang, his ad spend never got a chance, and no one will ever tell him the job existed. He did not lose the customer. He was edited out of the search.

I build software for contractors, so people expect me to say the robots are coming for the trades. They are not coming for the trades. Nobody is shipping a robot that can sweat a joint in a crawlspace. They are coming for the phone.

This is not a prediction. It is a shipping schedule.

Everything in that call record already exists in some form. Jobber sells an AI receptionist that answers calls and books work around the clock. Housecall Pro sells one too. A whole category of voice answering companies now sells to HVAC and plumbing shops for hundreds of dollars a month, and some of them, like Avoca, have raised serious money doing it. On the other side of the call, Google and the card networks are building a protocol that lets a person’s AI assistant buy things on their behalf, with real payments and signed proof of what the person authorized.

Meanwhile the thing being automated has not changed in twenty years. Harvard Business Review audited how fast companies respond to a lead and found the average response was 42 hours, and about a quarter never responded at all. Anyone who has called around for a plumber at 6pm knows the number that matters: most shops simply do not pick up, and almost nobody leaves a voicemail anymore. They hang up and call the next result.

So the pieces are on the table. Software that answers instantly. Software that shops and pays on a customer’s behalf. And an industry where the phone still rings out. You do not need to be a futurist to see what happens next. You just need to watch the pieces slide toward each other.

The three waves

The AI that answers your phone is about to be free

Right now, an AI that answers your phone and texts your leads back is sold as a premium. Hundreds a month, sometimes more, as its own product or as the top tier of software you already pay for.

That will not last. Answering, booking, and following up are becoming the entry fee, the same way a website went from “my nephew builds those” to something every business simply has. The models get cheaper every year, and every software company in the trades is racing to bolt the same feature on. What costs $400 a month today will be a checkbox in three years.

What it means for your shop: never sign a long contract for “the AI answers your phone.” You are buying a feature at its most expensive moment in history. Take the capability, sure. Just do not pay a premium for it, and do not let anyone tell you it makes them special.

The question flips from “what can it do” to “what did it earn”

Once every product answers the phone, the pitch changes. The question that will separate software worth paying for from software that is just loud will be brutally simple: what did it actually make me, in dollars, and can I check the math?

Think about how you treat an employee. You do not keep a tech because he describes himself well. You keep him because you know his callbacks, his tickets, his numbers. Software that acts on your behalf should face the same standard. If a tool claims it saved a job, it should show you the name, the date, the thread, and the invoice. If it cannot, it is telling you a story.

I think pricing follows the same flip. Software today charges per seat per month whether it earned anything or not. The moment a tool can prove what it earned, someone will offer to be paid on that instead, and every per seat vendor will have a very uncomfortable meeting.

What it means for your shop: start demanding receipts now. When a rep says their AI “books you more jobs,” ask to see how their product proves it, job by job, with names. Watch how fast the demo changes subject.

Your next customer might be a robot

This is the strange one, and the one I would bet on most.

Today, some of the calls coming into shops are already placed by software on behalf of some person or company. Tomorrow it gets personal: the same assistant that reads your texts and books your flights starts handling household problems. Water heater leaking. Furnace making that noise again. The homeowner will not want ten blue links. She will want it handled, and her assistant will call, compare, and book while she gets on with her evening.

When that happens, the shop that wins is not the one with the best ad. It is the one a machine can actually reach, get a straight answer from, and book. An assistant shopping for its owner does not wait through six rings, and it does not get charmed. It gets an answer or it moves down the list.

What it means for your shop: your phone manner will matter less, and your reachability and your reputation for straight answers will matter more. The shops that treat “a machine called and asked for a price window” as a real customer will quietly eat the market.

What to do about it

Whether or not you ever buy anything from me, here is what I would do if I ran a three truck shop:

1. Stop losing the calls you already get. Every wave above starts at the same place: the unanswered ring. Fix that first, with whatever tool you like. It is the cheapest revenue you will ever recover.

2. Demand receipts from your software. If it claims it earns you money, it should show the jobs, the names, and the dollars, and let you check. Anything that cannot is marketing with a login page.

3. Get your book in order. Your customer list, your job history, what equipment sits in which basement and how old it is. The machines work for whoever has the data. A shop with ten years of clean records will be worth more than one twice its size running on memory and a whiteboard.

4. Be reachable by machines. This sounds like science fiction until the first week it costs you a job. If an assistant cannot reach you and book you, it will book someone else, and nobody will ever tell you it happened.

The far end of this

Push the waves out a few more years and things get interesting.

Software you pay per saved job instead of per seat. Not a subscription you forget about, but a line item that reads like a commission for work you can verify actually happened.

An AI employee with a track record you can audit. Not a chatbot with a name, but something closer to a W-2: here is what it handled this month, here is what it earned, and here is what it got wrong, because a track record that never admits a miss is not a track record.

Equipment that reports its own failure before the customer notices. The water heater in that call record at the top should never have surprised anyone. The furnace knows it is dying weeks before the house gets cold. Someday the first “call” comes from the basement, not the homeowner.

And through all of it, one quiet constant: the answering bot becomes a commodity you plug in, and the record of your business, every customer, every job, every dollar, proven and portable, becomes the thing you actually own. Bots will come and go. The ledger is the asset.

Where I stand

I am building toward exactly this, from Champaign, for shops with one to ten trucks. I might be wrong about the years. I do not think I am wrong about the direction, because every piece of it is already shipping somewhere.

And I will hold my own software to the standard I just wrote: if it claims it earned you something, it shows you the receipt.

If you run a shop and want to argue with any of this, honestly, that is my favorite kind of email. And if you want to see what I am building about it, the demo is thirty minutes with me, not a salesperson.

Ian