Welcome back to The Blueprint, a regular feature where we highlight how Google Cloud customers are tackling unique and common challenges across industries using the latest AI and cloud technologies. We hope to inspire others looking to innovate in their work.
As the electric vehicle revolution gains steam in Europe, every charging session is a moment of truth, for the driver and for the transition itself. At lemonflow, we build AI agents purpose-built for EV charging: support agents for drivers and monitoring agents for charging networks. Our support agents answer drivers around the clock in more than 90 languages, and today they help an EV driver every 90 seconds.
Before we founded lemonflow, we worked closely with leading mobility players across Europe and saw first-hand how often charging goes wrong: Between 15% and 25% of EV charging attempts in Europe end in failure. Across Europe’s rapidly expanding EV market, where battery-electric car sales grew 30% in 2025 and thousands of charge point operators run public chargers, drivers who encountered stalled charging points were often left stranded or stuck in slow call center queues.
We set out to change that hands-on: to bring the cutting edge of AI to EV charging, and to build the world’s best conversational AI for distributed hardware, an automated platform that resolves real-time charging issues via phone, chat, email, and WhatsApp. However, handling live voice support created several technical requirements:
To deliver ultra-low-latency voice interactions and guaranteed serving capacity, we needed direct control over our model hosting environment.
We wanted every driver to get best-in-class support that feels effortless, almost magical: no queue, no menu, just a short conversation that ends with the car charging. Picture a driver on a long cross-border trip, pulling into a motorway service station at 11 p.m. She plugs in, but the charger won’t start: the previous session never closed properly. She calls the number on the charger and is answered in her own language. Within seconds, our agent spots the stuck session in the charger’s logs, restarts the charger remotely and stays on the line until power is flowing.
To get there, we had to solve our latency bottlenecks and secure dedicated compute capacity. We migrated from managed APIs to self-hosting the open Gemma 4 model on Google Cloud TPU v6e hardware in the EU.
Like many startups using Gemma, the compatibility of these open models with Gemini allowed us to transition our existing prompt infrastructure with minimal modification.
Moving to self-hosted TPUs unlocked the more complete architectural control our engineering team required to establish a deterministic core layer that handles EV-specific guardrails, guards against prompt injections, and standardizes previously fragmented backend protocols into our internal format before ever reaching the LLM. Control had a second payoff: because the model and serving stack change only when we change them, every test result stays meaningful. That stability laid the foundation for Agent Coach, our background coaching engine for AI agents, described below.
We then developed an intelligent routing gateway to prioritize workloads according to operational urgency, serving live voice calls first, followed by web chats, and deferring automated re-simulations to low-traffic periods.
Custom optimizations to the open-source vLLM serving engine and memory tiering between the TPU and host memory let a single TPU slice keep many more operator agents ready for instant answers, without recalculating their context.
We also implemented automated fallback routing to self-hosted GPUs and Gemini Enterprise endpoints. Thanks to TPU reliability, Gemma answers more than 99.9% of all live conversation turns, and these fallbacks serve primarily as deployment buffers during system updates. Primary or fallback, every request is served from secure EU compute.
For the e-mobility companies we work with, and for their drivers, the result is support that simply works: our agents typically answer every call, handle more than 80% of conversations end to end without a human handover, and earn 95% driver satisfaction. Behind the scenes, self-hosting Gemma on Google Cloud TPUs transformed our performance: With additional compute capacity, we introduced Agent Coach, a background engine that provides agents with relevant context during live calls and turns real conversations into permanent test cases, so every call makes the next one better. We also added nightly automated regression testing across thousands of fixed conversational scenarios.
Looking ahead, we are preparing to move to Ironwood, Google’s seventh-generation TPU, as soon as it is available in our EU region. With six times the high-bandwidth memory per chip, Ironwood will multiply the number of operator agents a single TPU can keep ready in memory for instant answers.
Ultimately, our goal is that no driver is ever left stranded at a charger. They get an instant, expert answer in their own language, the charger is rebooted remotely if necessary, and our AI monitoring agents catch the next fault before another driver ever notices. Open models on dedicated Cloud TPUs give us the speed and control to make that the standard across Europe.
Our platform runs two kinds of AI agents: support agents that talk to drivers in real time, served by Gemma on Cloud TPUs for minimal latency, and monitoring agents that analyze conversations and charger logs in the background on Gemini, so operators can fix faults before the next driver is affected. To keep its core platform consistent across many kinds of e-mobility companies, from charge point operators (CPOs) and e-mobility service providers (eMSPs) to charger manufacturers, lemonflow built a translation layer that converts varied CPMS interfaces into a standardized format.
When a driver calls the support line posted on a charger, the call reaches our Conversational Engine through the operator’s own hotline. The engine then pulls the charger’s live status and Open Charge Point Protocol (OCPP) logs, the messages chargers exchange with their backend, from the operator’s CPMS in real time. Because charging stations often lack digital displays, drivers are frequently unaware of underlying issues like pending mobile 3D-Secure payment authorizations, session timeouts, charging cards not enabled for roaming, power reduced by local load management, or a car that ended the session at its own charge limit. Our engine feeds structured log summaries directly into Gemma on TPU to inform interactions with drivers.
Direct model control allows us to stream token output selectively. Partial text tokens stream instantly to our Text-to-Speech (TTS) engine, while structured tool-call tokens stream asynchronously to backend systems to execute real-world actions, such as remotely restarting chargers, unlocking stuck cables, stopping sessions that failed to end, sending drivers details via SMS or email, or notifying service technicians.
Deterministic security controls screen inputs before they reach the TPU-hosted LLM, blocking malicious prompts. This architecture is designed with European critical-infrastructure requirements in mind, and it still keeps conversations snappy, with a typical response starting in around 350 milliseconds.
In the end, all of this should stay invisible to the driver: they just get a fast answer and a charger that works. We’re proud to work with Google Cloud to make that the norm and help EV adoption keep growing.