Skip to content
gitogi

AI integration

Six visits for one fault: the answer was in the files all along

Prokon halved technician visits: the answers to faults were already in its manuals. What this means for a maintenance business with ten technicians.

7 min read

This article was generated with AI assistance and may contain inaccuracies — verify critical information. How we use AI.

Six visits for one fault: the answer was in the files all along

On average, each wind turbine fault at Prokon required six site visits. Technicians would get in their vans and drive hundreds of kilometres. Travel time came on top of the work on site, leaving equipment out of action for long periods and the team worn out.

Yet the answers to those faults already existed. They were in an archive of 2,600 technical documents, amounting to more than 50,000 pages of documentation. The challenge was finding the relevant information and getting it to the technician before they set off.

Six visits for a single fault

Prokon is Germany’s largest energy cooperative and produces around 1% of the country’s electricity. If you maintain lifts, boilers or fire protection systems, you work on different equipment from Prokon’s turbines. But you may recognise the point where a job becomes difficult: the technician sets off without knowing exactly what to expect.

The story is told in a press release issued on 2 November 2026 by Prokon and its technology provider, octonomy, and in octonomy’s case study, which includes Prokon’s own account. After introducing AI assistants, Prokon halved the number of visits needed for turbine maintenance.

Twenty generations of turbines and scattered knowledge

Prokon operates turbines from twenty different generations, each with its own system and fault codes. Preparing for a visit means finding the documentation for the specific machine. It is a familiar challenge if you maintain equipment of different ages and have to track down the right manual each time.

The information needed was spread across documents, turbine monitoring data and the experience of second-line support staff: the experts technicians turn to when they do not know how to proceed. Reaching a diagnosis meant bringing together knowledge from different sources.

If you work in field service, you know the consequences of information arriving too late. A technician may only discover on site that they are missing a detail they need to proceed, forcing another visit. Prokon’s starting point was an average of six visits per fault.

How the assistant works, step by step

Prokon introduced an AI assistant: software that consults technical documentation and uses it to help identify the problem. Its work begins when a turbine sends a fault code.

The assistant compares the code with current monitoring data showing how the turbine is operating. It then searches the manuals and diagrams for that turbine generation. It can also interpret power curves: graphs showing the turbine’s power output. The people scheduling the visits and the technician receive the diagnosis before the technician sets off.

The assistant connects to the field service management software used to organise jobs and teams. The information therefore feeds into day-to-day work through tools already in use, without requiring a change in how activities are organised.

When the assistant does not know the answer, it says so rather than guessing. An uncertain answer presented as reliable could send a technician working on equipment in the wrong direction. Prokon’s head of IT and digital specifically highlighted the safety the system provides for people working on critical infrastructure.

Prokon has halved the number of visits, and problems are resolved 25% faster on site. The feasibility study took ten days, and the assistant was in use within six weeks. Those six weeks refer to bringing the assistant into use: the sources do not specify how long it took to halve the number of visits.

What this means for a business with ten technicians

Prokon operates on a very different scale from a small maintenance business, but the same problem can arise on a smaller scale. Consider a typical business with around ten technicians maintaining equipment of different makes and ages. This is an illustrative example to relate the case to everyday service work. Manuals are scattered across computer folders, ring binders and vans. Finding the diagram for the oldest piece of equipment means asking the colleague with years of experience working on it.

In that situation, your technician may end up calling that colleague from the customer’s car park. Or they may discover on site that the available diagram does not match the equipment and have to postpone the job. A second visit ties up a technician again, keeps the customer waiting longer and forces you to rearrange the schedule for other jobs.

A Gartner survey, presented at its customer service conference and reported by Customer Experience Dive on 4 November 2026, offers some guidance on where to focus your efforts. In Gartner’s survey of 275 customer service and support managers, only 4% of organisations had achieved the return they expected from AI: four in every hundred. The organisations furthest ahead invest almost 1.8 times as much in data and knowledge sources as in AI applications themselves. They also teach their people how and when to use the tools.

The survey covers customer service and support. The survey period is not specified. Even so, if you organise field service visits, it offers useful guidance: before introducing an assistant, you need to understand what information it will be able to consult and how usable that information will be. Prokon’s case shows what this looks like in practice: documentation already held in the archive helps prepare for a visit before anyone sets off.

How to get started

Start by choosing one team or one type of job. Defining the scope lets you examine a specific problem without trying to sort out all your company’s documentation at once.

Then review completed jobs. For faults that required more than one visit, work out where the information you needed was held: in a manual, a diagram or a colleague’s experience. This helps you distinguish time spent on the repair from time lost finding the information needed to prepare for it.

Next, take stock of your documentation. For every generation of equipment you maintain, check where the manual and wiring diagram are kept and whether they can be accessed on a computer or only on paper. Paper documents need to be made readable in digital form, because even an answer already written in a manual has to be retrievable by the assistant.

You can then work on connecting the assistant to the documents and your field service management system. From the outset, require it to flag when it cannot answer and deliver the diagnosis to the technician through the software they already use. Connecting AI to your company’s documents and software is what we do through our AI integration service: https://www.gitogi.com/en/servizi/integrazione-ai/

At Prokon, the feasibility study took ten days, and the assistant was in use within six weeks. To work out the timescale for your business, you need to review the material available and the work required to make it accessible. The inventory helps with that too.

How we work

At Gitogi, we help Italian maintenance businesses go digital and use AI through Odoo, AI integration and custom software. You can read how we work with technical service businesses here: https://www.gitogi.com/en/field-service/

Before recommending a solution, we want to understand how your team works and which steps cost them time. During the initial analysis, we work with you to choose a team or type of job to start with.

The first practical step is to review one completed month’s work, including that month’s job reports, invoices and contracts.

For the oldest piece of equipment you maintain, does your technician have the right wiring diagram before setting off, or do they have to track it down on site by calling the colleague who remembers it?

If that question brought to mind a fault that took more than one visit to fix, write to us and tell us how you work. We’ll start with one month of your service jobs, including job reports, invoices and contracts.

Frequently asked questions

How did Prokon halve technician visits?
It introduced an AI assistant that compares the fault code with current monitoring data and searches the manuals and diagrams for the relevant turbine generation. The people scheduling the visits and the technician receive the diagnosis before the technician sets off.
How many visits did each fault previously require?
Six on average, with technicians driving hundreds of kilometres. Prokon operates turbines from twenty different generations, each with its own fault codes.
Can a small maintenance business use an AI assistant like Prokon’s?
Scattered information can also be a problem in a small business that maintains equipment of different makes and ages. Start by choosing one team or one type of job, then check where your manuals and diagrams are kept and whether they are accessible digitally.
What happens if the AI assistant does not know the answer?
At Prokon, it says so rather than guessing. Make this a requirement from the outset, so your technician knows when the assistant cannot help.
Where do you start with Gitogi?
With one completed month’s work, including job reports, invoices and contracts.

Sources

  1. Windkraft-Journal, comunicato stampa di Prokon e octonomy AI GmbH «Prokon nutzt octonomy AI um Windkraft-Wartungsanfahrten zu halbieren»(accessed 9 November 2026)
  2. octonomy, caso cliente «How Prokon made critical infrastructure safer with AI»(accessed 13 November 2026)
  3. Customer Experience Dive (Matt Vartabedian)(accessed 9 November 2026)

Updates on ERP, software and AI

Practical insights on ERP, bespoke software and AI in business workflows. No spam.

Redazione Gitogi

Gitogi Srl

LinkedIn

Let's talk about your next digital step.

A free 30-minute call to see where your business can grow — with Odoo, AI and custom software. No commitment.

Free, with no obligation. Francesco replies within 24 working hours. Be ready to explain how you work today and what's getting in the way.

Related articles