Skip to content

AI · 6 min read

What artificial intelligence will change in asset management, what will stay the same and what the AI Act and Annex 22 say in 2026.

A chromatograph, an analytical balance and a microscope on lit pedestals. Each has a status alert, and all of them connect to a proposal waiting for confirmation.

Every few weeks a new AI model comes out promising to change everything. In asset management, some things are going to change a lot. Others will not change at all.

It pays to know which is which before you spend a single euro. This guide separates the two lists and reviews what the law says in 2026.

What AI will change

1. How you look up a fact

Today, finding out when an asset was calibrated means opening the inventory, searching email and checking a spreadsheet. With AI, you type the question in your own words and the answer comes with the document it is based on.

2. How you prepare for an audit

Preparing for an audit of your assets is mostly about gathering paperwork: certificates, inspection records, handover records. That is exactly the part AI can do for you, asset by asset, for a person to review.

3. How you keep track of due dates

A spreadsheet tells you what is due, if someone opens it. AI warns you in time and ranks the alerts: what comes first and why. Forty identical alerts do not help. Five ranked by risk do.

4. How you spot what does not match

AI can cross-check the official inventory against the last time each asset was seen and flag the differences: the asset listed in one room and scanned in another, or the one that has not been seen for months.

5. How you decide what to replace

Replacing by age is easy. With AI you can see usage, breakdowns, age and how critical each asset is in one view, and decide with better judgement.

6. Who does the preparation work

This is the most recent change: AI agents. An assistant answers. An agent prepares a complete task, such as opening a review of an asset that cannot be found or gathering its record, and leaves it pending until a person confirms it.

What AI will not change

1. Who answers to the auditor

In front of an auditor or an inspector, it is a named person with a defined role who answers. The AI Act requires human oversight for high-risk systems (Article 14), and applying the same standard everywhere else is good practice.

2. The need to calibrate

AI can warn you that a calibration is due. Calibration itself still takes a reference standard, a method and a technician. AI organises the work, but it does not replace it.

3. The importance of data

AI fed with doubtful data only produces more convincing errors. Without a unique identifier for each asset, a dated history and the documents attached to each record, any AI tool will fail.

4. The role of the official inventory

The ERP and the asset register remain the official record. AI reads them and flags differences. Correcting them is a person’s job, in the source system.

5. The duty to keep records

The obligations remain. Royal Decree (RD) 1215/1997, Spain’s work equipment regulation, for example, requires the results of work equipment inspections to be kept for the equipment’s entire working life. AI helps you find them, but someone has to have kept them.

The difference in one table

TaskTodayWith AIWhat stays the same
Finding a certificateFolders, email and spreadsheetsOne question and the linked documentChecking it is the right one
Preparing for an auditWalking the rooms and gathering paperworkRecords compiled asset by assetDealing with the auditor and signing
Due datesA spreadsheet someone has to openAlerts ranked by priorityDeciding whether to calibrate, replace or retire
InventoryA count once a yearDifferences flagged as they appearCorrecting the official record
ReplacementsBy ageUsage, breakdowns, age and criticality togetherDeciding the investment

What the law says in 2026

The AI Act. The EU Artificial Intelligence Act (Regulation (EU) 2024/1689) was amended this summer by the so-called Digital Omnibus on AI, Regulation (EU) 2026/1744, published in the Official Journal on 24 July 2026. These are the changes that matter most here:

  • Obligations for the high-risk systems in Annex III move to 2 December 2027.
  • Obligations for systems built into regulated products, in Annex I, move to 2 August 2028.
  • The AI literacy obligation (Article 4) stays in a revised form: those who use AI systems must take measures to support AI literacy among their staff.

It is worth checking case by case whether a specific use falls under Annex III. Applying human oversight even where it is not mandatory is the prudent choice.

Annex 22 of the EU GMP guidelines. It is the first annex of the EU Good Manufacturing Practice guidelines devoted to artificial intelligence. The draft provides for continuous oversight of AI systems, change control, model performance monitoring and human review when necessary. The public consultation closed on 7 October 2025 and the final text is pending publication.

Why most AI projects in asset management fall short

  • They start with the model instead of the data. Without organised data, AI has nowhere to get reliable answers from.
  • They do not cite their sources. An answer nobody can check is no use to an auditor, and the team stops trusting it.
  • They automate decisions a person should sign. And the first mistake costs the whole tool its credibility.
  • They try to cover every asset at once. The project drags on before it has proved anything.

How to prepare, step by step

  1. Give each asset a unique identity, the same in every system. See permanent identifier.
  2. Record every change with its date: moves, loans, calibrations, inspections and disposals.
  3. Attach every document to its asset, with its validity date.
  4. Require the AI to cite its sources and to say when information is missing.
  5. Write down what a person always decides before you switch the tool on.

Frequently asked questions

Will AI replace the calibration technician or the quality manager?

No. It will take search and preparation work off their hands. Calibrating, deciding and answering to an auditor remain jobs for people.

Do I need sensors to use AI in asset management?

No. AI works with the data you already have: the inventory, the ERP, the certificates and every label scan. What matters is that it is organised by asset and dated.

What is an AI agent?

An AI system that, as well as answering, prepares a complete task: gathering a record, opening a review or proposing a disposal. In asset management, the prudent approach is for it to leave each task pending confirmation.

Do I have to comply with the AI Act if I use AI for my assets?

It depends on the use. The obligation to support AI literacy among staff applies to anyone who uses AI systems. The high-risk obligations depend on whether the use falls under Annex I or Annex III of the Regulation.

How SEMBLA™ approaches it

SEMBLA AI is designed to answer with the data on your assets, cite the document behind every data point, say when information is missing and leave every change pending until a person confirms it.

We cover it in detail in SEMBLA AI: what it is, what it is for and how it works and on the SEMBLA AI page.

In short

AI is going to change how you work with your assets. Who answers for them will not.

Sources

  1. 01Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligenceEUR-Lex
  2. 02Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 amending Regulation (EU) 2024/1689OJ L, 24.7.2026
  3. 03Stakeholders’ consultation on EudraLex Volume 4: Chapter 4, Annex 11 and new Annex 22European Commission
  4. 04Real Decreto 1215/1997, de 18 de julio (Spain’s work equipment regulation), Article 4BOE, in Spanish
SEMBLA team · NEP TechnologyPublished 14 October 2026

Together, we will go through what SEMBLA would look like with your assets.