Skip to content

AI · 6 min read

What SEMBLA AI is and what it does for asset management: the questions it answers, how it cites its sources and why a person confirms every change.

A dark laboratory with an open laptop next to an analytical balance and a microscope, lit by the turquoise glow of the screen.

When was the analytical balance last calibrated? Where was it on 14 September? Which assets are due for calibration before April?

They are simple questions. Answering them is not. The data is spread across the inventory, email and several spreadsheets, and often only one person knows where to look.

SEMBLA AI is designed to solve exactly that. This guide explains what it is, what it is for, how it works under the hood and what it does not do.

What is SEMBLA AI?

SEMBLA AI is the artificial intelligence assistant in SEMBLA, the asset intelligence platform from NEP Technology. You type the question the way you would ask a colleague, and it is designed to answer with the data on your assets and link the document behind every data point.

It rests on three design rules:

  • It cites the source of every data point. Every answer links the certificate, reading or record it comes from, with its date.
  • It says when information is missing. If there is not enough data, it says so and points to the missing document.
  • A person confirms every change. SEMBLA AI prepares reports, alerts and proposals. Nothing is applied without sign-off from a named person with a defined role.

What it is for: 6 uses in asset management

  1. Finding a specific fact. ‘When was the analytical balance calibrated?’ The answer comes with the certificate linked.
  2. Knowing where an asset was last seen. ‘Where was the balance on 14 September?’ SEMBLA AI answers with the last dated signal: when, where and how it was seen.
  3. Anticipating due dates. ‘Which assets are due for calibration before April?’ Based on each certificate’s date and the calibration interval.
  4. Spotting what does not match. When the official inventory says one room and the latest record says another, SEMBLA AI flags the difference for a person to review.
  5. Preparing reports. Summaries of the state of your assets and of each one’s record, ready to review before an audit.
  6. Flagging gaps. ‘Which assets have no record of their last inspection?’ Knowing what is missing is information too.

How it works, step by step

Under the hood, SEMBLA AI follows a simple sequence:

  1. It understands the question. However you phrase it: ‘the balance in lab 2’ or ‘analytical balance’.
  2. It searches your data. It looks at each asset’s record, its history, its documents and the links between assets, locations and the people responsible.
  3. It answers only with what it has found. Every data point carries a number that links to its source.
  4. It warns you if something is missing. If the data is not enough, it says so and points to the missing document.
  5. It proposes and waits. If it suggests an action, such as opening a review, it leaves it pending until a person confirms it.

This method has a name: retrieval-augmented generation, or RAG. The idea fits in one sentence: search first, answer second.

An example from start to finish

Example with fictitious data.

Question: Where was the analytical balance on 14 September?

Answer: On 14 September 2026 at 10:42, the QR label of the analytical balance was scanned in Room 2.07 [1]. That is the last time it was seen. The official inventory places it in Room 2.05 [2]. The difference is consistent with the 2021 move [3] and is pending review by a person.

Sources: [1] QR code scan · 14 Sep 2026, 10:42 · Room 2.07 [2] Official inventory, record 2019/0388 · Room 2.05 · does not match [3] Signed transfer record · 3 Jun 2021

Notice three things. The answer does not claim where the balance is today; it says where and when it was seen. It flags that the official inventory does not match. And it corrects nothing on its own: it leaves the difference for a person to review.

SEMBLA AI vs a general AI chat: the real difference

A general AI chat is very useful for drafting or summarising. To answer questions about your assets, it lacks the essentials: your data and its history.

General AI chatSEMBLA AI
Where the answer comes fromWhat it learned online and whatever you paste inYour organisation’s data and documents
SourcesNot always givenNumbers and links every document
If it does not have the dataIt may fill the gap with something plausibleIt says so and points to what is missing
DatesIt does not know your assets’ historyIt answers by date: where an asset was on a given day
PermissionsIt sees everything you paste inIt answers only with what each user is allowed to see
ChangesIt has no access to your systemsIt proposes, and a person confirms

What SEMBLA AI does not do

  • It does not decide for you. It prepares the information and proposes. The decision and the signature belong to a person.
  • It does not fill gaps. It is designed to answer only with your organisation’s data and documents. When information is missing, it says so.
  • It does not monitor people. SEMBLA manages assets. People appear on the record as the ones responsible, never as subjects of tracking.
  • It does not replace your ERP or your official inventory. It reads them and flags the differences so they can be corrected in the source system. More on this in Integrations.

Your data: where it is and who sees it

  • In the European Union. Data is hosted on servers inside the European Union. More detail in Security.
  • Within each user’s permissions. SEMBLA AI answers only with the information the person asking can see on the platform.
  • No training of third-party models. The architecture is designed so that your organisation’s knowledge is not used to train third-party models without an explicit decision.
  • Not tied to a single model. AI models change every few months. SEMBLA AI is designed to switch models without touching your data or its history.

Why SEMBLA AI starts with the data

AI fed with doubtful data only produces more convincing errors. That is why SEMBLA AI is the last piece of the platform, not the first.

To answer well, it needs every asset to have a unique identity, a dated history and its documents attached to the record. It is the order we explain in Asset intelligence: what it is and why it is not the same as an inventory.

Frequently asked questions

Can SEMBLA AI make up an answer?

It is designed to answer only with your organisation’s data and documents, and to cite the source of every data point. If there is not enough information, it says so and points to what is missing.

Does it make decisions on its own?

No. SEMBLA AI prepares reports, alerts and proposals. No change is applied until a person confirms it.

What can I ask it?

For example, which assets are due for calibration this quarter, where an asset was on a specific date, who it was assigned to or which assets have no record of their last inspection.

Who can use it?

Any SEMBLA user, with the same permissions they have on the platform.

Do I have to change my ERP?

No. SEMBLA starts from the data in your ERP, your inventory or your maintenance software, reads it without modifying it and flags the differences.

What does the AI Act say about an assistant like this?

The EU Artificial Intelligence Act requires human oversight for high-risk systems (Article 14). SEMBLA AI applies that principle to all its proposals: a person confirms every change.

In short

SEMBLA AI searches, cross-checks and proposes. It cites the document behind every data point, says when it does not know and leaves the final word to a person.

You can see how it answers on the SEMBLA AI page.

Sources

  1. 01Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence, Article 14EUR-Lex
SEMBLA team · NEP TechnologyPublished 14 October 2026

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