Service Ticket
It's 7:15am and the service inbox has thirty unread: a web form, a voicemail transcript, a text the owner forwarded from his cell, two replies on a thread from March. This agent reads each one, finds the customer and the unit, checks whether the work is under warranty, on a plan, or a callback on your own crew, and opens the ticket with that answer and its evidence attached. It never dispatches, and it never decides what you can bill.
Yours outright — no required subscription.
Every agent is built for your business — your systems, your approval chain, your way of working. We scope it on the call.
If this sounds too technical, don't worry. We take care of everything for you.
- Private VPS deployment. This agent runs on a server that belongs to you, not a shared cloud tenant. Private VPN, firewall, and encryption are configured before it reads a single customer request, and nobody else's data touches your box.
- A technician agent alongside it. Every hire ships as a pair: the Service Ticket Agent doing intake and triage, and a VPS-technician agent that lives on the same box and keeps it healthy — patched, backed up, monitored. You're not hiring one thing, you're hiring a small team of two.
- Connection to every channel requests actually arrive on. Your shared mailbox, the web form, voicemail transcripts, the messages your owner forwards from his phone. Nothing has to be re-sent to it, and replies that land on an old thread get read too.
- Your triage rules, out of one person's head. During setup we write down what your coordinator knows: which accounts are on a plan, which carry a response clause, who's certified on what equipment, what counts as no-heat urgent. The agent runs those rules, and you can read and edit them.
- Coverage checks against your own records. It reads the serial and model off the request or the customer's photo, matches it to your equipment and warranty records, and reports what it found with the source behind it. Where it can't verify, it says so instead of guessing.
- A live walkthrough of both agents, so you see exactly what it opens a ticket for, how it writes a coverage line, and where it's supposed to stop. Plus 14 days of priority support after launch, to tune it against your real request mix.
- You own everything — the server, the agent, the code it runs on.
Not sure it fits? Check fit in 90 seconds in the free assessment chat.
Who this is for
- You're running an 18-person shop where every service request lands in one shared inbox, and one person retypes the address, the description, and the model and serial into your ticket system before anyone gets dispatched.
- You serve a mix of contract and break-fix accounts, and every request has to be checked against that customer's agreement before a tech touches it. The hours nobody caught as billable are the leak you can't see.
- You sell and service equipment, so the first question on every call is whether the unit is still under manufacturer warranty and whether anyone actually registered it, and the answer lives in a distributor portal plus a spreadsheet.
- You've had a request come in Friday at 5:10pm and sit untouched until Monday at 7, and you've had the one person who knows how to triage call in sick. On both days, service didn't get triaged and you heard about it from the customer.
How it earns trust
The failure mode of AI in service intake isn't that it gets a coverage answer wrong. Your coordinator gets one wrong too. It's that you can't tell which ones. This agent makes the opposite bet: every ticket it opens shows what it read, what it matched, and what it decided not to touch.
Every field traces back to a sentence
Each ticket carries a link to the exact inbound message it came from, and each field shows the quoted source line it was filled from. You can see whether "no cooling, upstairs unit" was read out of the customer's words or inferred by the agent.
Coverage is written as a claim plus its evidence
Which serial string it read, which record it matched, which install and registration dates it found, and which source it read them from. When a lookup fails it names the lookup that failed, rather than defaulting to covered or not covered.
A daily digest, including the no-ticket list
It lists every message it processed, including the ones it decided were not service requests. That list is where a missed job hides, and it's the one you read for the first two weeks. Nothing is removed from your inbox — the originals stay unarchived, so the shared inbox remains an independent source of truth to reconcile against.
How much of your intake it clears depends entirely on how clean your equipment and account records are, and we'd rather size that with you on the call than promise a number here.
Pairs well with
These share a workflow with this role. Tick any to add them to your setup.
Half her morning is copying emails into the ticket system. Does this actually stop that?
It clears the routine ones. A plain request with a readable address, a described symptom, and an account it can find comes through as an opened ticket with the fields filled and the coverage line attached. The messy tail still lands on her: a serial it couldn't match, an account it couldn't identify, a suspected duplicate. It flags those rather than forcing an answer, so the honest expectation is that her morning gets shorter, not that it disappears.
We ate the labor on a unit that turned out to be under warranty the whole time. Will it catch that?
It catches the ones your records can answer. It reads the serial, matches it to your equipment records, and reports warranty, plan, and callback status with the evidence behind it. Two real limits: if coverage data lives in a distributor portal it can't log into, it returns unknown and hands the ticket to a human, and a transposed character in a serial can match a real record and produce a confident wrong answer. That's why it shows the raw string it read next to the record it matched, so a bad match is visible before anyone drives out.
What decisions does it make on its own?
Fewer than people expect. It never dispatches or schedules; who drives out and when stays with whoever runs your board today. It never makes a coverage decision that costs money — it reports status with evidence, but it does not approve a claim, waive a diagnostic fee, or mark work non-billable. It never files a claim with a manufacturer or distributor, and never quotes a price, a trip fee, or an arrival window. Outbound to the customer is limited to the acknowledgment wording you approve.
A request came in Friday night and nobody saw it until Monday. Does that stop?
The intake part does. It reads what arrives when it arrives, opens the ticket, and sends the acknowledgment you approved, so the customer isn't sitting in silence and Monday starts with a triaged queue instead of a pile. It does not wake anyone up or send a truck. And urgency lives in tone as much as in words: "my mother is on oxygen" is a P1 written calmly, so it triages by the rules you give it and will miss urgency your rules never encoded.
Nobody knows who owns a ticket. How does routing work?
It routes off the roster and the rules you give it, and every assignment shows the rule that fired — commercial account, no-cool, mini-split certification, therefore this tech. That means a bad assignment traces to a stale rule instead of a black box. It does not learn interpersonal history: the tech who left in March, the account that won't accept a particular person, the certification that lapsed. Someone has to tell it, and updating the rules is a five-minute job you own.
What's covered under the $1,997 setup, and what isn't?
One company, one ticket system, one escalation path. It's a flat one-time setup, scoped on a free 15-minute call where we look at how requests arrive today, where coverage data actually lives, and who triages when your coordinator is out. Typically about seven days from kickoff to real requests running through it. Multiple ticket systems or genuinely different triage policies per division get scoped on that call, not priced off this page.
Hiring more than one? A department on tap — the subscription puts a build team behind every request, agent after agent.