Using AI on your energy data without the risk

Energy data has always belonged to the few people trained to read it. Until now. The AI in EG EnerKey opens it to everyone else in the organization – the controller who needs the figures for a report, the facility manager who wants to know why a building spiked last night, the new colleague analyzing consumption without waiting for the energy specialist.  

Opening data to more people only works on one condition: it must stay protected. 

Using public LLMs for energy data analysis: The high-risk shortcut  

The most common way AI enters energy management today is through a spreadsheet. Export the metering data, paste it into a public chatbot, ask a question. It is fast. But it moves regulated operational data into a service nobody in IT has assessed, with no record of what was shared. 

Netskope's 2026 cloud threat report found that incidents of sensitive data sent to AI applications doubled in a year, to an average of 223 per organization per month. And since 2 August 2026, the EU AI Act's transparency rules apply, and the energy sector sits on NIS2's list of essential services. Meaning the tools you use are your responsibility as well as the vendor's. 

The question is not whether to use AI on energy data or not, but how to do it without taking on a data risk. Iiro Kuntsi, Product Manager for EG EnerKey, has spent the past year building an AI just for this. 

What “safe” AI means in practice for energy management 

The interesting part of the AI-agent in EG EnerKey is not the chat window. It is what sits behind it.  

The AI runs inside EG's own EU cloud tenancy, on the same infrastructure that already hosts EG EnerKey. There's no new environment for a customer's IT or security team to review and approve. It authenticates as the logged-in user rather than through a separate account. It can only access the facilities and data that the user's role already permits, nothing wider. It's built on an enterprise language-model service, not a consumer AI product, which makes all the difference. Prompts and outputs are never used to train public models, and nothing is sent to a public AI service. Every query runs through a controlled layer with explicit, engineering-approved access to EG EnerKey's own functions, and every interaction is logged for audit.

And, the AI is read-only: it can analyze but cannot change anything in the platform. EG is a European company operating under European rules – GDPR, the AI Act, NIS2. Under the AI Act the assistant is classified as limited-risk, a decision support rather than an autonomous decision-maker, with answers traceable to the data they came from. That is what the Act's transparency and human-oversight duties ask of you as the organization using it. Iiro continues: 

Data residency and security haven't come up as objections. That doesn't mean they're unimportant. It means we designed the AI so the answers to any concern are already there.

Security and compliance: how the AI in EG EnerKey handles your data → 

Five questions to ask before using AI on your energy data 

1. Where is the data processed, and under whose law? EU region, and a European company accountable under GDPR and the AI Act, should be the baseline. 

2. What can it see? It should inherit your existing user permissions – not a service account with broader access. 

3. Can it change anything? For analysis, read-only is the safe answer. 

4. Is it logged? Every question and every data retrieval should be logged in a form your security team can audit. 

5. Is your data training a public LLM model? The answer should be no. 

If all five answers are good, the shortcut stops being tempting – and the people who were exporting spreadsheets get a better tool than the one they had. 

Which brings us back to where we started. Open energy data to more people. Keep it as secure as it always was. Build the tool together with the people whose problems it solves. When those three hold, the controller, the facility manager and the newest colleague can ask their own questions about the buildings – without waiting for the specialist to have time, and without anyone exporting a spreadsheet. 

Already an EG EnerKey customer? The AI can be added to your existing plan – contact your account manager.

Insights on energy management and AI

Related products