Get in, set your destination, drive off – even longer journeys are no real problem with modern EVs. The car calculates necessary charging stops automatically, and with Plug&Charge, you just plug in on site – or pay briefly via app or card. Every year, more charge points go live and gaps in the network are shrinking.
Still, the situation is far from ideal – and it is not about the debate over charging prices, ad-hoc payments, or roaming. Users are understandably frustrated when the car or charging app shows a functioning, free 300 kW charge point, but in reality, the session won’t start, the display is broken, or the promised 300 kW drops to just 30 kW. A quick 20-minute stop can easily turn into a longer delay, and detours to other charging parks cost extra time.
That is where Evailable comes in. The company is part of E.ON and wants to bring the group’s experience from energy networks – where uptime is critical and AI is already established – to e-mobility. We spoke with Maren Rehnelt, Managing Director of Evailable, about what the company promises, how it works, and how failures can be detected up to a week in advance.
Evailable wants to make charge point availability a given. How does your “real availability” differ from the familiar uptime of a charge point?
I believe both EV drivers and operators experience the difference every day. Uptime means the station is online. Real availability means the charging session is successful. Our system goes beyond simple availability reporting. We analyse errors, charging behaviour, communications, and many other factors. One example: a station appears as “available”, but charging sessions have been failing or only running at half power for two weeks. Evailable makes such issues visible – classic backends do not.
Let’s say the AI tool detects a fault. What happens next? Is there a standard message sent to the charge point operator? Or does the system trigger a response before sending a notification? Often, a simple restart is enough.
It depends on the problem. In some cases, a restart definitely helps. Our AI has learned when this is useful and performs it immediately. Across all customers, we help over 100,000 times per month quickly and automatically, without operator intervention. For other issues, we show the operator the factors that led to the problem.
It’s a bit like a doctor: with Evailable, the operator sees both the patient’s history and the diagnosis, including all symptoms. The right solution can be applied immediately – without lengthy manual analysis. That is our goal: automate, save time, save money.
How does Evailable calculate the “Health Index” that indicates the charge point’s condition? If there is no fault yet, can none be detected? Do historical comparisons suggest that a component is likely to fail soon?
We approach the Health Index similarly to medicine. Early warning signs can appear long before a major failure. We analyse the full raw data stream from connected stations and compare it with other stations of the same type – every model behaves slightly differently. Our neural network is trained for this. We can predict failures up to a week in advance.
Which data points feed into the evaluation? Are basic data enough, or does more data improve the prediction?
In most cases, we work with already available OCPP communication data. This allows operators to start with Evailable immediately. Additional data – such as SIM info or sensor data from individual components – make our predictions even more precise, down to component-level forecasts.
How is the data captured technically? Is additional hardware required, or does the AI tool use existing interfaces and sensors? Are there differences between charging point models?
The big advantage: you don’t need any extra hardware. Configuration or data connections on the stations don’t need changing. It’s straightforward. In most cases, the OCPP raw data sent from the station to the backend is simply forwarded to us, so we can perform our analysis. Interfaces with many common CPO backends and proxy solutions already exist, keeping the effort minimal. Differences aren’t in data collection but in interpretation – our AI adapts model-specifically.
Who exactly in the charging value chain is Evailable aimed at?
Our solution is used by three types of companies. Firstly, CPOs who want to optimise networks, reduce operating costs, and increase successful sessions. Secondly, backends that integrate our technology via API to bring AI capabilities into their platforms. And finally, charge point manufacturers looking to improve service and hardware performance efficiently.
How did the idea for Evailable come about? Who is behind the company?
The idea emerged years ago in E.ON’s innovation division, led by my colleague and fellow MD Stefan Herr. Two worlds collided: scaling issues in charging infrastructure and predictive maintenance knowledge from electricity and gas networks. The prototype was tested internally and performed so well that it spun off into its own company. That’s how Evailable was born. Today, we help CPOs in 14 countries make charging reliable.
This interview was first published in German and has been translated. It is part of the media partnership between Ev Authority and the Intercharge Network Conference (ICNC), taking place 2–4 September in Berlin. Maren Rehnelt is one of the speakers at ICNC25. Her talk is scheduled for 3 September at 12:40.
