AI automation for service companies from the job call to the invoice
This page is for a company that does the work at the customer's site or in its own service shop: installations, inspections, maintenance, repairs, fitting crews, cleaning and technical facility upkeep. The job call, the visit, the service report and the invoice are one chain, and today it breaks in four places. We stitch it back together and give technicians an AI agent that answers from your equipment documentation.
- 14 daysto the first working automation, counted from the end of the workshop
- EUR 3,500 netper week of workshop, a published price we do not negotiate
- 5 to 10 secondshow long the AI agent takes to answer a technician standing at the unit, from your own documentation
The problem
Why does the invoice go out three weeks after the job is done?
The job call arrives on the owner's mobile, by email to the office inbox and through the form on your website. Three channels, three lists and no shared queue. Who drives out and when is decided in the dispatcher's head and passed on by text, so the only place the whole day's plan exists is his memory.
Then comes the visit, and that is the moment the data stops existing. The service report stays on a sheet of paper in the van, the photos stay on the technician's phone, the parts used get written down in the evening from memory or not at all. The office does not know what was actually done at the customer's site, so it has nothing to raise an invoice from and waits for somebody to bring the paper in.
That slippage runs against a hard deadline on the other side. Tax law sets a date by which the invoice for a completed service has to be issued, counted from the month the work was done, and it is measured in days. A service report coming back after three weeks eats almost all of that slack, and on a job finished at the end of the month it eats it down to zero. Your accountant knows the exact date in your country; what matters here is that the clock starts when the technician leaves the site, not when the paper reaches the office.
Service contracts and periodic inspections sit apart from all of this, tracked in a spreadsheet one person keeps. As long as she remembers, everything works. When she forgets, the reminder comes from the customer or from an inspector.
If nobody in your company drives out to a customer and you sell expert time from behind a desk, this page is next to the point. You are closer to the pages for marketing agencies, IT companies, training companies, construction companies, food service or hospitality.
- A job call from the inbox gets lost, because the queue exists in three places at once
- The technician drives out a second time, because he had no part with him on the first
- The invoice waits for a service report that sits in the van until the end of the week
- An inspection from a service contract drops out of the spreadsheet and the customer reminds you
Why now
What changed, so a technician's phone is enough to replace a system?
A mobile form stopped being an IT project. The technician opens a link on his own phone, takes one photo, fills in three fields and collects the customer's signature with a finger on the screen. You do not have to buy a separate field service management system, and you do not have to hand the crew tablets, to get the data from a visit back the same day.
The second change is on the accounting side, and it decides the schedule. A lot of service companies run their accounting on a server in their own office rather than in the cloud. A local program has no address the internet can send an event to, so you reach it through a relay running inside your network, for example an n8n instance on your own machine. That is a craft with a known pattern today, not research. Cloud billing tools such as Fakturownia connect over a plain API, and so does Make if you would rather keep the orchestration off your premises.
The third thing is the AI agent. A rule will carry a job number from one system into another, but it will not answer a technician standing at the unit who needs the torque figure for this model and the procedure behind a specific error code. An AI agent reads your documentation, equipment cards and service procedures and answers in 5 to 10 seconds, instead of sending him off to call a colleague who is on another site.
What changes
What does the office get, and what does the technician get?
One queue for every job call
Phone, email, web form and chat land on a single list with a priority and a response time taken from that customer's contract. The dispatcher sees what is urgent before he starts calling people back.
Response time read from the customer contract, not from the dispatcher's memory
A visit confirmed with the customer
The schedule is built on crew availability, and the customer gets the slot confirmed the moment the job is assigned. A moved visit goes out to the customer on its own too.
Confirmation at the moment of assignment, not a phone call the evening before
The service report from a phone, not from paper
A photo, a short description, parts used picked from a list and the customer's signature on the screen. All of it created at the unit, not in the evening from memory.
One photo and three fields, filled in on site at the customer
The invoice built from the report
Data from the service report goes straight into your billing tool or into the accounting program sitting on the server in your office. The office stops chasing the technician to find out what he actually did.
Everything the invoice needs is ready on the day the job was done
An inspection that reports itself
Service contracts and periodic inspections live in a calendar together with the date from the contract. The reminder reaches the dispatcher with a lead time you set yourselves.
Contract dates watched by a calendar, not by a spreadsheet
An AI agent for the crew in the field
The technician asks in a plain sentence about an equipment parameter, an error code or a procedure, and gets the answer from your own documentation. No ringing round colleagues.
An answer from company documents in 5 to 10 seconds
Proof
Do you use this yourselves?
Yes, and we will say straight away what this proof does not cover: our own service business sells expert time and sends nobody into the field. What is shared is the office half, the road from job call through settlement to invoice. Below is half a year of production measurement in our founders' company, together with what those numbers do NOT prove.
Built for ourselves. These are not client deployments but implementations inside the founders' own companies. We say so plainly, because proof you cannot verify is not proof.
JustBottheir agentJustIdea Agency, a performance marketing agency of around 30 people
- 23
- people from the company work with the agent
- 38
- systems wired in permanently
- 112
- processes run without a human
- 475
- thousand lines of the agent’s code
The agent is wired into Google Ads, Meta, the CRM, the task system, the SEO tool, Analytics, Search Console and the data warehouse. First commit 1 March 2026, 23 people use it today.
8 of 38 systems it works in
Google Ads
Meta Ads- HubSpot
- ActiveCollab
- Ahrefs
GA4
Search Console
BigQuery
- Before
- In the twelve months before launch the agency ran on a margin of minus 4.42 percent. Across a twenty month window there was not a single run of three positive months.
- After
- May, June and July 2026 were three positive months in a row, the first such run in the whole window. Revenue barely moved, so the change sits on the cost of delivery, not on sales.
Margin up 5.55 percentage points year over year, on revenue up 1.5 percent
Operations run by the agency agent in consecutive months of 2026, from the first commit in March. From 83 operations in March to almost 63,000 in August, with users growing from 4 to 23. This curve is here because a rollout nobody uses looks exactly like a successful one in a report.
Measurement: Agent invocation log in the data warehouse, March to August 2026, counted at the level of a single tool call.
In your company
What if our people never touch it?
Four sentences come up most often in service companies. All four are fair, and we answer them here the way we would answer them on a call.
Our technicians will not fill anything in.
That is often true, and training will not change it. Which is why the form from a visit is one photo, three fields and a signature with a finger on the screen, less work than the sheet of paper somebody has to carry back to the office. If even that does not stick, the whole office side still stands: the job queue, the schedule, inspection reminders and an invoice built from data the office holds without the technician. That is a smaller scope and a smaller quote, not a failed rollout.
Our accounting software sits on a server in the office, it will not connect.
It will connect, just differently from a cloud tool. Programs like Comarch Optima or Subiekt have no address the internet can send an event to, so we run a relay inside your network and it talks to the program locally. We check this during the workshop, because the scope and the schedule depend on it. Cloud billing tools go over a plain API and take days rather than weeks.
We have six technicians, that is too few to roll anything out.
With six technicians it is usually not worth buying a separate field service management system, and we will tell you so. It is worth having one job queue, a service report from a phone and an inspection calendar, because those three cost the same at six people as at twenty. If after the workshop we do not see even that, we say plainly that there is nothing here to automate.
Our job calls live in email and on a phone, not in any system.
That is the starting point in most companies we talk to, and you do not have to change it before we start. The inbox and the phone stay as channels, because the customer will call anyway. What changes is that both land on one list with a priority, instead of in two separate heads.
- CRM and customers
- HubSpot, Pipedrive, Salesforce
- Job call channels
- Website form, company inbox, phone, chat
- Work in the field
- Mobile forms, phone photos, on-screen signature
- Invoicing and accounting
- Fakturownia, cloud billing tools, Comarch Optima, Subiekt
- Team work and data
- Google Workspace, Microsoft 365, Slack, Google Sheets
- Automation layer
- Make, n8n, custom integrations over API
What exactly we automate in a service company
The usual split of work between a job call and a paid invoice, before the rollout and after it.
| Area | Today | After the rollout |
|---|---|---|
| Job calls from customers | phone, email and form, three separate lists | one queue with priority and the response time from the contract |
| Assigning a technician | worked out in someone's head and sent by text | a schedule built on crew availability, with the slot confirmed |
| Service report | a sheet of paper in the van, or nothing | a form on the phone: photo, description, parts, signature |
| Parts and materials used | written down afterwards from memory | picked from a list at the customer, straight onto the job |
| Invoice | raised once the paper reaches the office | everything it needs is ready on the day the job was done |
| Inspections and service contracts | a spreadsheet one person keeps | a calendar that reminds you before the contract date |
| Questions from the field | a call to a colleague or to the office | an AI agent answers from documentation in 5 to 10 seconds |
The scope of the first rollout is set during the workshop. We take the three items with the highest return, not all of them at once.
How we work
Five steps from paper in the van to an invoice the same day
Every step ends with something you can judge before you decide on the next one.
Workshop
One week. We walk the route of a job call from the customer's phone through to a paid invoice and mark every point where data has to be retyped by hand. We also ride along on one visit, because the list of things to fix looks different after a day in the field than it does from the office.
Picking three automations
From the map we take the three with the highest return and measure them before the start: how many days pass between the job being done and the invoice, how many job calls get lost in a month, how many inspections were done late.
First rollout
Fourteen days to a working solution. In a service company we usually start with the report from the phone and the data behind the invoice, because that is one change you can see in the bank account in the first month.
Running it in the field
Two weeks on real jobs. We watch which fields the technicians skip and why, then cut those fields instead of reminding the crew about their duty. The form adapts to the visit, not the other way round.
Measurement and the next one
After 30 and 90 days we check the indicators agreed at the workshop. Only then do we decide whether to take another area or whether what already runs is enough.
Before and after
How the dispatcher's week changes
Before
- Monday goes on copying job calls from three places onto one list
- Visits worked out in someone's head and sent by text, every change is a round of calls
- Last month's invoices wait for service reports coming back from the vans
- The customer reminds you about a contract inspection, not the calendar
After
- Monday starts with a ready queue, with priority and response time on it
- The visit has a slot confirmed with the customer at the moment of assignment
- Everything the invoice needs is ready on the day the job was done
- The inspection reports itself, with the lead time written into the contract
The number of visits does not change, and neither does repair time. What changes is the number of days between the job being done and the invoice going out, and whether the contract inspection happens at all.
For the owner of a service company
What the person signing this off gets out of it
The first thing shows up immediately. The question of what exactly was done at that customer in March stops requiring a call to a technician who is away on holiday, because the answer sits on the job together with the photo and the customer's signature. The same goes for a complaint, where today you lose on one person's word against another's.
The second is less obvious, and it matters most in a conversation with a bank or when you sell the company. The service history of the equipment and the delivery of your service contracts stop being knowledge held by a few people and become data you can pull from a system. The business stops depending on who happens to remember.
- The road from job done to invoice raised is shorter and measured in days
- Service contract delivery is visible before the date, not after it
- The history of a unit stays in the company when the technician leaves
- Scope and price for the rollout are known after the workshop, before you decide
What it costs
The workshop has a published price, the rollout is quoted after it
The process workshop costs EUR 3,500 net per week and we give you that number before we talk. In a service company one to two weeks is usually enough, because the process is long but narrow: from job call to invoice. The rollout is quoted after the workshop, once we know whether the accounting sits in the cloud or on a server in your office. That is a difference in the work, not a line in a price list.
Tools and models you pay for directly with the provider: Make, n8n, your billing tool, your language model provider. Model queries run at roughly EUR 0.005 to 0.07 each depending on the model, and it goes on their invoice, not ours. No commission from us and no reselling.
Process workshop
EUR 3,500 net
per week, one to two weeks for a typical service company
A closed piece of work. It ends with a map of the job call route and a 30 day plan you can carry out without us.
- A week of work in your company, in the office and on one field visit
- A map of the road from job call to paid invoice
- Three automations with the highest return, counted before the start
- Indicators to measure after 30 and 90 days
- Most popular
Workshop and first rollout
Quoted after the workshop
scope follows the number of systems and where the accounting sits
The variant most companies pick. The report from a phone and the data behind the invoice run within 14 days of the workshop ending.
- Everything from the workshop
- One job queue and a visit schedule with customer confirmation
- A service report from the technician's phone, with photo, parts and signature
- A link to invoicing, including a program on the server in your office
- A calendar of inspections and service contracts with reminders
- Training for the person who runs this on your side
Service plus an AI agent for technicians
Quoted after the workshop
for companies whose crew calls in for knowledge several times a day
Everything from the rollout plus an AI agent that answers from equipment documentation and service procedures.
- Everything from the first rollout
- An AI agent answering from equipment cards, manuals and procedures
- Access from a phone in the field and from a computer in the office
- Alerts on job calls past their response time and on inspections before the date
If after the workshop we decide that tidying up the inspection calendar and one report template is enough, we say so plainly and show you how to do it. The workshop ends with a document that stands on its own without us and that you can carry out with your own hands.
Frequently asked questions
What does automating a field service company actually mean?
It means a system takes over the steps somebody does by hand today between a job call and an invoice: collecting calls from several channels onto one list, scheduling visits, writing the service report, carrying it into the invoice and watching the inspection dates in service contracts. The data usually exists already, except that part of it lies on paper in a van and on a technician's phone. Automation is about creating it in a usable form from the start.
Do the technicians have to use a new app?
They do not have to install anything separate. The form from a visit opens from a link in the phone browser: a photo, three fields, a list of parts and the customer's signature with a finger on the screen. If the crew still will not touch it, we build the office side only, which means the job queue, the schedule, reminders and invoicing from data the office holds without the technician.
Will you connect to accounting software running on our own server?
Yes, but differently from a cloud tool. Local programs such as Comarch Optima or Subiekt have no address the internet can send an event to, so we run a relay inside your network and it talks to the program on site. We check the version and the access method during the workshop, because the schedule depends on it. Cloud billing tools connect over a plain API.
How long does a rollout take and where do you start?
One week of workshop, then 14 days to the first working automation. We usually start with the report from a phone and the data behind the invoice, because that change shows up in the bank account in the first month. The job queue and the inspection calendar come next. We measure the effect after 30 and 90 days, on indicators agreed before the start.
Will this shorten repair time or bring us more jobs?
We promise neither, because repair time depends on the equipment and on parts availability, and the number of jobs depends on the market. We promise two things you can check: everything the invoice needs is ready on the day the job was done instead of after the paper reaches the office, and an inspection from a service contract does not fall out of the calendar. The rest we measure after 30 days and show you without dressing it up.
How do you keep track of inspections and service contracts?
Dates from the contracts go into a calendar together with the unit, the site and the person responsible. The reminder goes out with a lead time you set yourselves, and an inspection that was not done stays on the list instead of disappearing once the date passes. Contract delivery becomes a number you can see before the date, rather than after a customer complains.
What exactly does the AI agent do for technicians?
It answers questions from your documentation: equipment cards, manufacturer manuals, service procedures, the repair history at that customer. The technician asks in a plain sentence from his phone and gets an answer in 5 to 10 seconds. The agent does not guess: if something is not in the documents, it says it does not know instead of inventing a parameter.
How much does the rollout cost after the workshop?
It depends on the number of systems we connect to and on whether the accounting sits in the cloud or on a server in your office, because the second case is more work. We give the quote after the workshop, once we know the answer. The workshop itself costs EUR 3,500 net per week and you know that price before the first conversation.
Let us walk through one of your job calls together
The free consultation takes 30 minutes. You walk us through one job call from the customer's phone to a paid invoice, and we tell you where we see days to recover and where there is nothing worth automating. You do not have to give us access to any system at this stage. You write directly to Pawel, no reception and no tickets.
We run at most two rollouts at a time, so that somebody is genuinely sitting on each one.
Book a free consultationPS. If you raised an invoice late this month because the service report was lying in a van, that is exactly the problem. After the rollout the invoice data exists on the day the job was done, and the contract inspection reports itself.