Zendesk for EV charging support teams: drivers, site hosts and station data in one system

The problem: a driver at a charger that will not start, at night, in a language your agents may not speak

EV charging support is support for an asset in the field, with a customer standing next to it. Drivers ask why a session will not start, why they were charged twice, or how to use the app. Site hosts ask about hardware faults and uptime. Questions arrive in English, French and Spanish through a mobile app, a website and email, and most of them repeat. Meanwhile agents look up station records by hand because the charger database and the help desk have never been connected.

How we approach Zendesk for EV charging

An AI agent inside the mobile app

Drivers are already in your app when they need help. For a global energy company's EV charging business we configured a [Zendesk AI agent](/solutions/zendesk-ai-agents-implementation) connected to a Zendesk messaging channel built on the Zendesk messaging Mobile SDK. The client ran a proof of concept inside its app, and we reviewed live conversations to find where answers needed tuning. The chatbot went from proof of concept to live conversation review in about five months, with an expected 30 percent of repeat charging questions answered before they reach an agent.

Multilingual conversation design

For a charging network software company we configured a Zendesk AI agent that answers in English, French and Spanish, with dialogues for payments, hardware and site issues, the app and general questions, and a persona with short, direct answers. Knowledge sources were the help center, the company website and uploaded data. Payment and hardware problems escalate to people.

Payment lookups on escalation

A driver who was double-charged does not want to read an article. During payment escalations the agent looks up customer details and charges in Stripe, so the person who picks up the ticket already has the transaction in front of them. A separate agent handles ticket email, and a voice channel can follow.

Station records on the ticket

Agents should never look up a charger by hand. Using Polytomic, we synced station and location data from the client's MongoDB database into Zendesk custom objects. When an agent enters a Chargebox ID on a ticket, Zendesk triggers link the ticket to its station and location records. The same design works for any charger management system with an exportable database or API.

Measuring the bot and the agents separately

Chats the AI agent resolves collect a star rating. Tickets handled by agents use a separate survey, so the network can compare the two and see where the agent needs tuning without blending the numbers.

What to check in your own setup

• Where drivers ask for help first: app, website, phone or the sticker on the charger.
• Which languages your drivers and site hosts use, and whether your help center exists in each.
• Where station, site and session data lives, and whether it can be synced or queried by charger ID.
• Which payment processor holds the charge records and whether it has an API for lookups.
• Whether drivers in your app are authenticated, so escalations can carry identity into the ticket.

Buy Zendesk through ShadowRock

ShadowRock is an authorized Zendesk reseller and a Zendesk Premier Implementation Partner. Charging networks and charging software companies can buy Zendesk Suite and AI agents from us with licensing and implementation in one contract. If you are moving from another help desk, our [migration page](/solutions/migrate-to-zendesk) covers the cutover. See [how buying Zendesk through ShadowRock works](/partners/zendesk/buy). Verify our Zendesk partner status on the Zendesk Marketplace.