The agent explains the trade-off and collects the inputs. It does not make the recommendation — your technician does, after seeing the equipment.
An AI agent for repair versus replace conversations is a 24/7 digital assistant that explains the trade-off framework your company uses, collects the equipment facts that framework needs, briefs the technician, and books the visit where the decision is actually made. It stops homeowners walking into a five-figure decision in a hot house with no idea how the math works and every reason to suspect an upsell. The agent teaches the method, not the answer. It walks through the factors you weigh — age against expected service life, the cost-versus-age rule you apply, refrigerant type, warranty status and the efficiency difference — and gathers each input from the customer. Then it says, in plain words, that the recommendation belongs to the technician who will see the unit. That boundary is deliberate and it is not adjustable by anything the customer says.
Explains the framework, gathers the inputs, hands the decision to a person.
Explains the rule your company uses — the $5,000 rule, or whatever threshold you actually apply
Collects the inputs the rule needs: age, repair history, refrigerant type, warranty status
Describes what each factor does to the math without reaching a conclusion
States plainly that the recommendation is the technician's to make, not the agent's
Sends the completed input sheet ahead of the visit so nobody starts cold
Books the diagnostic or the comfort consultation, whichever your process calls for
This is where home services sales goes wrong most often. The customer suspects an upsell, the technician has forty minutes to explain thirty years of equipment economics, and a large decision gets made by someone who is hot, tired and already out of pocket. Moving the explanation earlier changes the dynamic entirely. When the homeowner already understands why a sixteen-year-old R-22 system is a different conversation from a six-year-old unit under warranty, your technician confirms and prices a decision instead of arguing for one.
A simple, three-step flow.
The agent walks the homeowner through the factors your company weighs — expected service life, your cost-versus-age rule, refrigerant type, warranty status and the efficiency difference — and is explicit that it is describing the method, not applying it.
It collects the unit's age from the data plate, the refrigerant type, what has been repaired in the last two years and what the current failure looks like, confirming each answer back to the customer.
The completed sheet goes out with the job. The technician sees the unit, makes the call and prices it. The agent has done the explaining; the licensed person does the deciding.
A realistic use case with concrete timing and output.
Scenario: A 6-technician HVAC company runs about 20 repair-or-replace conversations a month through August and closes roughly a third. On a Saturday afternoon a homeowner texts that the outdoor unit is running but the house is at 82. The agent gets the model and serial off the condenser, decodes a 2009 build, identifies R-22, and learns the capacitor was replaced last summer and the system has been topped up with refrigerant twice since. It explains what each fact means — R-22 has not been produced since 2020 so a recharge is expensive, two top-ups in two years points to a leak, and sixteen years is at the end of the expected service life — then says clearly that the technician will make the recommendation on Monday. On Monday the conversation takes 25 minutes rather than an hour, because the homeowner has already talked it through with her husband.
Roles that gain practical value from this AI agent.
You arrive at a conversation the homeowner has already had with themselves, instead of starting the education at minute one.
The framework has been explained by someone with nothing to gain, which is worth more than any presentation you could give.
Standardizes what the company says about repair versus replace instead of leaving it to each technician's version.
The most valuable conversation in the business stops depending on you being free to have it.
Gives the office an honest, safe answer to the question they are least equipped to handle.
The same age-against-repair-cost math applies to a corroding tank or an ageing boiler.
Key tools and what the AI agent does inside each.
Runs the explanation as a text conversation the homeowner can re-read and forward to their partner.
Reads the equipment record and prior job history, then writes the completed input sheet to the new job.
Holds your thresholds, expected-life table and serial decode rules, so the framework stays yours.
Attaches the collected inputs to the work order the technician opens on arrival.
Logs the conversation against the contact so the replacement follow-up has the history behind it.
Six practical scenarios that this AI agent excels in.
Common questions about using the AI agent in workflows.
An AI agent for repair versus replace conversations is a 24/7 digital assistant that explains the trade-off framework your company uses, collects the equipment facts that framework needs, briefs the technician, and books the visit where the decision is actually made. It stops homeowners walking into a five-figure decision in a hot house with no idea how the math works. Unlike a voicemail box or a booking widget it holds a genuine conversation about the trade-off — and unlike a salesperson, it never makes the recommendation itself.
No, and that is the central design decision on this page. The agent explains the factors — age against expected service life, the cost-versus-age rule you apply, refrigerant type, warranty status, the efficiency difference — and gathers the inputs. It does not weigh them and it does not produce a verdict. The recommendation comes from a licensed technician who has seen the equipment. If the customer asks "so what should I do", the agent says that is the technician's call and books the visit.
It is a rough industry heuristic: multiply the age of the unit in years by the cost of the proposed repair, and if the result clears $5,000, replacement is usually the better value. Plenty of companies use a different threshold or ignore it entirely. The agent uses whichever rule you configure and describes it as a rule of thumb rather than a verdict, because on its own it ignores refrigerant type, warranty status and how well the unit has been maintained.
It gives your published replacement range for the equipment class and tonnage they describe, says clearly that a firm figure needs a load calculation and a look at the ductwork, and books the appointment. It will not narrow the range to make the customer happy and it will not put a number on a system nobody has sized. Holding that line is the reason to script this conversation rather than leave it to whoever picks up the phone.
It works the other way round. The agent gives the homeowner the framework and the facts about their own equipment, then says the technician decides. When your technician recommends replacement and the reasoning matches what the homeowner already learned from a source with nothing to sell, the recommendation reads as consistent rather than opportunistic.
It walks the customer to the data plate and asks them to type the model and serial. Your serial decode rules, held in a sheet you control, turn that into a manufacture year, and the model identifies the refrigerant. If the plate is unreadable or the serial matches no rule you have supplied, the agent records what it has and leaves the age blank rather than guessing a year.
Yes. The shape of the decision is the same for water heaters, boilers and heat pumps: expected life, repair history, warranty status and the efficiency gain from replacing. You supply an expected-life table and a threshold per equipment class, and the agent runs the same explain-and-gather flow without ever crossing into the recommendation.
The agent explains the trade-off and collects the inputs. It does not make the recommendation — your technician does, after seeing the equipment.