AI knowledge base for plumbing and building services (HVAC): the practical guide

For plumbing and building services businesses – heating, air conditioning, ventilation and sanitation, known in Austria as HKLS – an AI knowledge base makes scattered knowledge about installations and experience findable by asking in plain language. Instead of searching in folders, manufacturer portals and the heads of individual technicians, the team asks, for example: “Which settings were documented for the heat pump at Gartenstraße?” – and receives the answer together with the source document. The biggest lever is not invoices or other accounting documents, but documented know-how: commissioning reports, service reports, checklists and tried-and-tested solutions to problems.

In brief

• Plumbing and heating businesses look after installations for 15 to 25 years – the documentation has to outlast any single employee’s time with the business.
• The most valuable knowledge often sits in the heads of a few experienced technicians and disappears when they retire or change jobs.
• An AI knowledge base answers questions from your own documents in plain language and always names the source document.
• Customer service and the emergency service benefit most: anyone standing in front of an unfamiliar installation needs reports and fault history within minutes.
• Quotes, invoices and stock levels stay in the ERP or the industry-specific software; the knowledge base takes care of the how and why.

Why is knowledge so scattered in a building services business?

Because it comes from many places at once and is rarely brought together. The office writes quotes and order confirmations, the technician fills in service reports on paper or on a tablet, and manufacturers supply manuals and error code lists as PDFs or behind portal logins. And the crucial details – why the flow rate on a particular installation was deliberately set higher – are often not written down anywhere.

Every business knows the typical situation: a customer calls, the heating has failed. The colleague who has looked after the installation for years is on holiday or has left the business. The technician who goes out knows neither the configuration of the installation nor its history and starts from scratch – even though the business may already have fixed the same fault on the same installation twice.

Then there is the service life of the installations. A heating system runs for 15 to 25 years; during this time, employees, software systems and filing structures change several times. Paper folders from the time of installation, an old “Customer service” folder on the server and the memory of the longest-serving technician are then the only sources – and none of them is available when the emergency call comes in on a Sunday evening.

Which documents and knowledge belong in the knowledge base?

At the heart of it is everything that describes how an installation is built, how it was set up and how the business works. For a building services business, this typically means the following:

Knowledge	Typical content	Everyday benefit
Installation records per customer	Commissioning reports, hydraulic diagrams, settings, controller parameters	The technician knows the installation before setting off
Maintenance and service reports	What was done when, which parts were replaced	Fault history of an installation at a glance
Manufacturer documents	Manuals, error code lists, installation instructions	Look up error codes without portal logins and searching
Internal instructions and standards	Heat pump commissioning checklist, underfloor heating flushing report, brine circuit filling	Every technician works to the same standard
Resolved faults	Symptom, cause, solution, installation affected	Recurring problems are only solved once
Correspondence about installations	Agreements, change requests, complaints	Commitments remain provable, even after years

The focus is deliberately on instructions, reports and standards – not on accounting documents. A delivery note matters to the accounts department; the commissioning report with the settings actually used matters to the technician tracking down a fault three years later. This is precisely the knowledge that most businesses manage worst, because it has no fixed place in any of the existing systems.

What questions does the team ask day to day?

The benefit of an AI knowledge base shows in the questions that come up constantly in day-to-day business and currently lead to phone calls, queries or searching through folders:

1. “Which heat pump is installed at the customer’s in Gartenstraße, and which settings did we document at commissioning?”
2. “What does error code F28 mean on this combi boiler model, and how did we solve the case last time?”
3. “How do we commission a heat pump – which checklist currently applies?”
4. “What was done at the last service at the hotel on the edge of town, and which parts were replaced?”

The AI answers such questions from your own company documents and names the source document for every answer; one click opens the original. This matters because AI can make mistakes – the source makes every answer quick to check.

Text recognition (OCR, the automatic conversion of photographed or scanned text into searchable text) also opens up the world of paper: the signed service report, the old folder of commissioning reports, the installation logbook with handwritten additions. Systems such as junova, the AI knowledge base from Goma-it combine this text recognition with automatic pre-filling: the system reads the customer, installation, date and topic from the document and fills in these values itself. What remains to be confirmed is the document type – a person makes this decision in the review area, because the document type determines who gets to see the document later. For a clear-cut service report, that is one click.

What stays in the ERP and the industry-specific software?

The division of labour is clear: the ERP knows what happened – the knowledge base knows how it is done. Quotes, order confirmations, invoices, stock levels, time recording and job scheduling stay in the ERP or the industry-specific software; these systems are built for this and remain the systems of record.

The knowledge base sits one level above. It answers the questions for which there has been no system so far: how is the installation configured? Why was it set up this way? How did we solve this problem last time? Where accounting documents and knowledge are connected – for example when the report and the correspondence with the customer belong to the replaced circulation pump – it adds the context without replacing the accounts.

What is a realistic way for a plumbing and heating business to get started?

Start small and concrete, not with a big digitisation project. This sequence works well:

1. Collect search questions: For one week, note what the office, customer service and the emergency service actually search for. This list determines what goes in first.
2. Define structure and rights: Filing per customer or installation, plus an area for internal standards. Cost calculations and purchasing terms get their own access rights, so that not every technician’s tablet sees everything.
3. Load the starting set: First the active maintenance customers – reports, settings, service reports – and the internal checklists. Older records follow later, selectively.
4. Test with real questions: For two weeks, two or three employees ask their everyday questions and check the answers against the source documents.
5. Build it into everyday work: From a set date, new service reports and commissioning reports go straight into the knowledge base – photographed on site, not weeks later from the pile in the office.

The guide AI knowledge base in your company describes in detail how these steps work together across industries. Keep your expectations realistic: the first weeks take some effort for structure and the initial import; the benefit grows with every documented job.

The next step is manageable: take a look at how the introduction works step by step – from the intro call and the structure through to checking it with your own installation data. A business with well-documented maintenance customers usually has the most important documents together faster than the team expects.