This article is the first part of The AI Shift series, an editorial deep dive into how EV charging operations are being reshaped as the industry moves toward AI-native systems. It builds on conversations with some of AMPECO’s trusted partners in the EV driver support niche and combines their perspective with our view on what this transformation looks like in practice.
Norway is one of the world’s most mature EV markets, and the drivers calling for support look nothing like they did a few years ago. It’s no longer the tech-savvy Tesla early adopter who troubleshoots independently. Instead, it’s the first-time EV owner who doesn’t understand why the charger won’t start or the driver stranded at 10pm in the rain, growing increasingly frustrated, calling in a language your support team doesn’t speak.
“You’re getting the grandmas, you’re getting the granddads, the people who have no idea how to solve the problem,” says Andreas Helland, CEO of Atender, which provides scalable, multilingual, people-first customer support and CX solutions for companies across Europe. “That’s something a lot of CPOs are not really planning for.”
Every EV market follows the same trajectory as it matures: the audience broadens, the questions multiply, and the patience for poor support shrinks. The traditional approach to driver support, where you hire agents, train them, scale the team as your network grows, was never designed for the complexity this industry demands. AI doesn’t just make that model faster. It replaces it entirely.
The scaling math that doesn’t work
Here’s a reality that many CPOs underestimate: benchmarking across industries shows that EV charging generates a higher volume of support requests per customer than most comparable sectors. Public EV charging still produces double-digit failure rates. According to J.D. Power, 14% of EV drivers have shown up to a charger and couldn’t charge at all and ChargerHelp!’s 2025 Reliability Report confirms that nearly 1 in 3 attempts fail on the first try. Each of those failures is a potential support call. Why is this the case?
EV charging sessions involve hardware, software, payment processing, roaming agreements, grid conditions, and vehicle compatibility, all interacting in real time. When any link in that chain breaks, an EV driver calls.
Unlike many support scenarios, the phone remains the primary support channel because these calls are urgent. A driver stuck at a charger isn’t browsing a help center for answers; they need a resolution now. The urgency is physical, immediate, and tied to their ability to get home.
Now layer on what happens when you expand across markets. It means new languages, new hardware models with their own quirks, new roaming partners, and different cultural expectations for how support should sound and feel. A German caller expects formality, while a Dutch caller expects directness. A Finnish caller wants you to get to the point immediately. An American caller expects an apology before you even start solving the problem.
Hiring and training human agents to cover all of this, across time zones, at the quality level drivers expect, is prohibitively expensive. As Helland puts it bluntly:
“The EV industry is still working toward profitability, which creates a massive drive to handle support more efficiently.”
What “good support” actually means for EV drivers
Before talking about solutions, it’s worth defining what EV drivers actually need from support, because it’s different from what most industries optimize for.
Speed matters, but context matters more. When a driver calls about a failed session, the most frustrating part isn’t waiting on hold; it’s having to explain the situation from scratch, then repeat it when they’re transferred. The best support interactions start with the agent already knowing which charger, which session, and what likely went wrong.
Then there’s the language question, but not just whether someone can communicate in the driver’s language, but whether the interaction feels culturally natural. The difference between functional translation and native-feeling support is the difference between tolerable and trustworthy. And trust matters enormously when someone is relying on your infrastructure to get from A to B.
Finally, the real standard to aim for is making support unnecessary. The best support experience is one that never happens because the issue was detected and resolved before the driver ever noticed.
“Electricity is electricity,” says Zsolt Máté Juhász, CEO of United Call Centers, which delivers AI-driven, multilingual customer support, sales, and automation services to help companies scale customer experience operations worldwide.
“You cannot say that your electricity is better than others. So service is the main differentiator, and customer care is the heartbeat of it.”
How AI changes the model entirely
AI in customer support often gets reduced to “chatbots that answer FAQs.” That’s a feature, not a transformation. The real shift is structural, and it happens across three dimensions.The insights below come from a panel on the future of EV driver support at the Intercharge Network Conference (ICNC) Berlin 2025. Watch the full session.
1. From reactive to proactive
The most significant change AI enables is eliminating the need for responses altogether. When AI can correlate charger telemetry, OCPP logs, and historical support data, it can identify patterns that predict failures before they affect drivers. A charger showing early signs of a communication fault can be remotely reset overnight, so the driver who uses it the next morning never encounters a problem.
This only works when the EV charging management platform surfaces the data cleanly. Every OCPP event, session outcome, and hardware state needs to be accessible; that’s the foundation that proactive AI tools read from and act on. AMPECO’s CoOperator does this natively, auto-generating root cause analysis for hardware faults, failed sessions, and authorization failures so operators can act on issues before a driver ever encounters them. “Focus on operations first,” advises Patrick Roelke, CEO of AmpAssist, which delivers a complete AI Operations-as-a-Service platform that monitors, resolves issues, supports drivers, and optimizes charging networks 24/7.
“Focus on operations first. Prevent the calls from even happening. Run a better product, better UX, more transparency in pricing.”
Roelke sees support as a backstop for what operations can’t catch, not as the front line. AI makes that philosophy achievable at scale.
2. From linguistically limited to culturally native
AI voice and text agents can now operate in more than 30 languages, but the more interesting development is cultural adaptation. Konstantin Huneke, Co-Founder and CEO of Lemonflow.ai, startup that provides a specialized, multilingual customer support and operations platform for electric vehicle charging networks, describes building region-specific AI agents with distinct personalities: “In the US, the AI agent apologizes twice before getting to the point. In Finland, we get directly to solving the issue. In France, it’s more formal. In the Netherlands, it’s more informal.”
This level of localization would require enormous, specialized human teams to replicate — and even then, consistency would be nearly impossible to maintain across shifts and turnover cycles. AI delivers it as a configuration decision, not a staffing challenge.
3. From scaling headcount to scaling intelligence
Across the industry, practitioners consistently estimate that roughly 80% of EV driver support interactions are repetitive and follow predictable patterns: the charger didn’t start, the payment didn’t process, the session ended unexpectedly. These are exactly the interactions AI handles well and consistently, instantly, and at any volume.
This doesn’t eliminate the need for human agents, but it does transform what they do. Instead of handling the same five questions hundreds of times a day, your human support staff become specialists who can tackle genuinely complex cases with better tools and more context than they’ve ever had. AI provides them with parsed OCPP logs translated into human-readable summaries, full session histories, and probable root causes before they pick up the phone. Huneke reports that this approach reduces resolution times for complex cases by 20–25%.
How AI-powered support actually gets adopted
One of the biggest misconceptions about AI in EV driver support is that it’s a binary switch: either you’re running human support or you’ve “replaced” it with AI. The reality is much more gradual. As Huneke describes:
“Don’t think about AI as all or nothing. See it more as a transition you make step by step.”
He describes a common adoption pattern: CPOs start with after-hours support or a single language they can’t currently cover with human agents. They see results such as fewer missed calls, faster resolution, lower costs, and they expand gradually, from 10% of volume to 50% to eventually the full front line.
The practical starting point is the platform a CPO already runs on. AI support tools connect to the charging management system to pull session data, access OCPP logs, and trigger remote actions — that integration is what makes them effective, not the AI itself. Platforms with open APIs and pre-built support integrations shorten this path considerably.
AMPECO’s marketplace includes direct connections to both Lemonflow and AmpAssist, which means CPOs can activate AI-powered support without custom development work. Those on more closed systems will need to solve that problem before any of the tools above become practical
The advantage that compounds
There’s a strategic dimension to this transition that goes beyond cost savings. Helland makes this point forcefully: adopting an AI-centric approach to support delivers data insight that “will give you so much insight into improving the service outside of just your customer support, which is extremely valuable.”
He’s right. Patterns in support data reveal hardware issues before they show up in monitoring dashboards. Clusters of similar complaints expose UX problems in the app or payment flow. Regional trends highlight where network expansion or partner renegotiation is needed. Operators who adopt AI-centric support now are building an intelligence asset that grows more valuable with every interaction. Those who wait are behind on efficiency and on accumulated learning.
What happens to your support team
AI doesn’t eliminate the need for humans in support, but it changes their job description.
The agents who previously spent their days handling repetitive, often frustrating interactions move into a specialist role. They handle edge cases that genuinely require human judgment, armed with better context and diagnostic tools than they’ve ever had.
As Helland notes, the experience of being a support agent actually improves: “You have a much stronger basis to answer the customer on, and you’re more capable of solving the issues.”
Juhász goes further, predicting that “human touch will become a prestige, a premium service” — a white-glove tier for high-value customers or genuinely complex situations. Whether or not that prediction holds, the direction is clear: smaller, more skilled teams doing higher-value work, supported by AI that handles the volume.
The shift at a glance
The transformation in EV driver support isn’t a single change — it’s a fundamental restructuring across every dimension of how support operates. Here’s how the model shifts:
| Dimension | The Old Playbook | The AI-Native Future |
| Staffing | Human agents, limited hours (8am–6pm) | AI agents available 24/7 across all time zones |
| Response Time | Common queries take hours or days | Routine queries resolved in seconds |
| Scale | Teams grow linearly with customer base | Unlimited simultaneous conversations |
| Knowledge | Quality varies by agent; high turnover | Consistent expertise; continuously learns |
| Languages | Limited to 1–3 languages your staff speaks | Instant support in dozens of languages |
| Proactivity | Purely reactive—customers must contact you | AI proactively notifies users of issues |
What this means for your operations
EV driver support is not a function you grow linearly with your network, but it is a system you can redesign around AI before it becomes a constraint on your growth.
Operators who make that shift earlier rather than later will reduce costs, resolve issues faster, prevent more of them from happening, and build a support experience that keeps up with the pace of their expansion.
The rest will keep hiring and falling behind.
