At Kaluza’s latest Breakfast Club in London, leaders from energy and applied AI came together to discuss one of the most live questions in the sector: what does it take to become an agentic energy company?
The discussion brought together perspectives of leaders from E.ON Next, Faculty AI, Amazon and Kaluza, covering customer service, operating models, flexibility, data foundations, trust and the role of AI agents in the future energy system.
In the discussion, “agentic” did not mean using AI as a better search tool or adding a large language model to existing processes. It meant systems that can act with a degree of autonomy: interpreting context, choosing the next step, using tools or data and completing a task within clear guardrails.
What emerged was a grounded view of agentic energy. This is not about adding AI to every process or chasing the latest technology trend. It is about using intelligence, automation and context to make energy companies better at what customers already need from them: simpler experiences, faster support, lower costs and more value from a flexible energy system.
The technology is moving quickly. But the harder work is knowing where to apply it, how to build trust, and how to make increasingly complex energy systems feel simple for customers.
Start with the energy problem, not the AI
One of the clearest points from the discussion was that an agentic energy company is still, first and foremost, an energy company. The starting point is not “where can we use AI?” It is “what does the business need to do better for customers, colleagues and the energy system?”
Customers do not want AI for its own sake. They want an energy provider that is easier to deal with, more responsive, more affordable and better able to help them navigate a changing system.
That distinction matters. Agentic systems should not become a new layer of complexity for customers or employees. Its value is in bringing together the context energy companies already hold across customers, devices, meters, markets, tariffs, operations and service interactions, then using that context to make better decisions.
The opportunity is not simply to automate existing processes. It is to help energy companies understand customers more completely and act more intelligently on their behalf.
The best starting point is focused use cases
The conversation also challenged the idea that companies need a wholesale transformation programme before they can begin.
Energy companies are complex. Data is often fragmented, operating models are evolving, and legacy systems remain part of the reality. But waiting for perfect foundations risks paralysis.
A more practical route is to start with focused, high-value use cases: automating high-volume back-office tasks, helping service agents respond faster, improving billing workflows, supporting EV charging queries or surfacing better customer insights.
These use cases help organisations learn where AI works, where it does not, what guardrails are needed, and where humans still need to remain in the loop.
Not every AI use case needs to be agentic. Some problems are best solved by deterministic workflows, some by LLM-enabled support tools and others by agents that can act autonomously within defined limits. The real skill is knowing which tool to apply to which problem, and where autonomy creates enough value to justify the added complexity, cost and governance.
Trust is the operating model
If AI is going to act on behalf of customers, trust becomes central.
That is especially true in energy. Customers may be willing to let a provider optimise their EV charging, manage a battery or recommend a tariff, but only if the service works reliably and reflects their real needs.
A single poor experience can undermine confidence. A customer who is recommended solar panels for a home they rent, or whose car fails to charge when they need it, will not care how advanced the underlying model is.
This is why personalisation matters. Energy companies need to use the data and context available to them responsibly, so recommendations and actions feel relevant, useful and safe.
It is also why humans will remain essential. Vulnerable customers, customers in distress, and situations involving risk, confusion or complaint will still require human judgement. The future operating model is not one where people disappear. It is one where people are freed up to focus on the moments that matter most.
For customer, complexity needs to disappear
The future energy system will be more complex behind the scenes.
EVs, heat pumps, batteries, solar, smart tariffs, OEM platforms, network constraints, flexibility markets and regulatory obligations all need to work together. For customers, however, that complexity cannot become the experience.
The role of agentic systems is to abstract it away.
That could mean helping an energy specialist understand why a smart charger is not behaving as expected. It could mean coordinating millions of distributed assets in response to grid signals. Or it could mean identifying the best tariff or action for a customer without asking them to understand every variable in the system.
The customer should not have to know how many parties sit behind the service. They should simply feel that it works.
The market needs intelligence it can operationalise
Agentic energy is still early. There is no universal playbook for applying agents across energy businesses, and the technology is not suitable for every use case.
But the direction is clear. As the energy system becomes more distributed, dynamic and data-rich, companies will need to make decisions with more context and at greater speed.
The winners will not be those that simply “add AI”. They will be the companies that understand where autonomy creates value, where deterministic workflows are enough and where human judgment still matters.
Agentic energy is not about making energy feel more automated. It’s about making energy feel simpler.
