Digital maintenance instructions: on your phone instead of in a folder

Maintenance instructions are most useful where the work is done – on your phone in front of the installation, not on the folder shelf in the office. For this to work, you need three things: the documents in digital form, central storage and a search that understands questions in everyday language. This guide shows step by step how companies bring their manuals, service reports and installation data from paper to the phone – and where the honest limits are.

In brief

• Maintenance instructions only deliver their value when they can be accessed on site – in an office folder they help nobody standing in front of the installation.
• Most manufacturers’ manuals have long existed as PDFs; as a rule, a company only has to scan its own and older documents.
• Text recognition (OCR) makes scanned folders, photos of rating plates and old reports searchable too.
• An AI-supported knowledge base answers questions in everyday language and names the source document for every answer.
• For safety-relevant steps, the manufacturer’s original remains authoritative – the AI answer is the quick way there, not a substitute for it.

Why are maintenance instructions usually where nobody needs them?

The typical picture in technical companies: in the office there is a wall full of folders – manufacturers’ manuals, installation documentation, inspection reports, neatly labelled and grown over decades. Meanwhile, the fitter who needs the information is standing twenty kilometres away in the customer’s boiler room, puzzling over an error code.

What happens next is familiar to every service company: the call to the office. There, someone leafs through the folder, photographs the right page and sends it as a message to the phone. That works somehow – but it ties up two people every time, takes a quarter of an hour and fails completely if nobody in the office happens to pick up.

Even companies that are “already digital” are often only halfway there. The PDFs are on the server, spread across folder structures that only the person who created them understands. But storage is not yet findability: anyone who doesn’t know the storage location and file names will search in vain on a small phone screen.

Which documents belong in the digital maintenance file?

The digital maintenance file is the complete, searchable collection of all documents for an installation or an installation type – from the manufacturer’s manual to your own service note. Five groups cover everyday work:

Document	Typical content	Typical question on the job
Manufacturers’ manuals	Operation, maintenance intervals, error codes	“What does error code E42 mean on this boiler type?”
Your own service reports	Work carried out, parts replaced, anything unusual	“When was the circulation pump last replaced?”
Installation documentation	Rating plates, wiring diagrams, settings	“What values was the burner set to?”
Inspection and acceptance reports	Measurements, conditions, defects found	“What was flagged at the last inspection?”
Spare parts and supplier lists	Part numbers, sources of supply, delivery times	“Which valve fits, and who can deliver it fastest?”

The most valuable are usually your own reports and notes – because they exist nowhere else. The manufacturer’s manual can be obtained again if necessary; that the shut-off valve on the installation on the third floor sits behind the panelling is known only to the colleague who installed it.

That is exactly why a pure collection of manuals falls short. Only the combination of manufacturer knowledge and your own knowledge of the installations answers the questions that really come up on the job.

How do folders and PDF collections get into a searchable form?

The route from the folder shelf to a searchable maintenance file is more manageable than it sounds – provided you take it in the right order:

1. Review the collection: get an overview of which installation types you look after and which documents exist for them – on the shelf, on the server, in mailboxes.
2. Obtain manufacturers’ PDFs: most manuals from the last fifteen years are available as PDFs from the manufacturer. Downloading is faster than scanning.
3. Digitise your own material: scan service reports, installation sheets and important older documents; text recognition (OCR, the automatic conversion of images into searchable text) turns them into queryable documents.
4. Bring it together centrally: store everything in one place, named traceably by installation type or customer – instead of scattered across drives and mailboxes.
5. Make it queryable and check it: add the collection to an AI knowledge base and check with real questions from everyday jobs whether the answers and their sources are correct.

What this build-up looks like in practice in various companies – from the initial filling to everyday routines – is shown in the practical guide to the AI knowledge base.

What does querying in everyday language on the phone deliver?

The real difference shows on the job. The fitter types into the phone the question he would otherwise have phoned through to the office: “Which maintenance intervals apply to the type X ventilation system?” or “What was unusual at the last service of the installation at the lakeside hotel?” The answer comes quickly – formulated from your own documents, with the source document stated.

This source is more than a convenience feature. An AI can be wrong, so every answer must be verifiable: a tap opens the original document at the right passage. For safety-relevant information – pressures, torques, isolation procedures – this look at the original is mandatory, not optional.

In everyday work, the effect adds up: fewer calls back to the office, shorter jobs, fewer second trips due to missing information. And new fitters can ask questions without tying up an experienced colleague for every task. How such a knowledge base works – with data stored on EU servers and without your documents being used to train third-party AI models – is described on junova’s product page.

Where are the limits of the digital maintenance file?

Honesty is part of it: the technology makes documents findable, but not better. An outdated manual remains outdated, an incomplete service report remains incomplete – maintaining the collection remains work that the company has to organise.

Text recognition has its limits too. Printed manuals and clean scans work reliably; scrawled handwriting on yellowed installation sheets, on the other hand, is often only partly recognised. For such cases it has proven useful to briefly add the key information when capturing – one line of context turns a silent scan into a findable document.

And finally, reception: in a basement without a signal, the best database is of no help. For plannable jobs in dead spots, you download the relevant documents in advance – a step that belongs to job preparation just like the tools.

How do companies get started without a major project?

The most common mistake is trying to digitise thirty years’ worth of folder wall in one go – that overwhelms any team and fizzles out after three weeks. The opposite has proven itself: start with the twenty installation types that generate the most questions, and expand the collection bit by bit from day-to-day business.

This way the benefit arises first where it saves the most time – and the team experiences early on that capturing is worthwhile. How the introduction works in practice, from the initial conversation and the initial filling to the first real questions on your own documents, is shown in the implementation process for the AI knowledge base.