More than a regular chatbot
IGP Agents are much more than a chatbot or a fancy search function: they can understand and carry out complex tasks. The agents can interact with one another, which allows them to work across multiple areas of the Intelligent Grid Platform. They can access the same functions available to users – from power flow simulations and a wide range of analysis tools to the grid planning toolbox.
Because the AI behind IGP Agents is not simply added on top of the Intelligent Grid Platform as a chatbot, but deeply embedded within it, users spend less time searching for information, switching between systems, and piecing together individual steps. The result is faster, more efficient, data-driven decision-making.
Security comes first
Grid operators work with sensitive critical infrastructure data that must be protected. That is why data from the IGP is neither shared with commercial AI providers nor used to train their models. Two core principles provide additional safeguards for IGP Agents:
- Human in the loop: IGP Agents can propose analyses, studies, or actions, but cannot execute anything without explicit approval from a user.
- Data minimisation by design: Personal data is removed or anonymised before being passed to the language model, regardless of user permissions, while technical grid data is retained.
In addition, the underlying language model has strictly limited, read-only access and runs in a sovereign European environment where data is not stored.