How Electra's customer care team handles complex support tickets 80% faster
- Industries
- TechnologyOther
- Company Size
- 201-1000
- Department
- Customer Support
Key Highlights
- 80% reduction in time spent resolving escalated customer tickets
- 3 specialized support AI agents handling invoices, refunds, and complex inquiries
- 4 new AI conversations created per hour for customer care agents
About Electra
Electra builds and operates fast-charging stations for electric vehicles across Europe. With over 300 employees spread across 10 countries, the company has positioned itself as a leader in the EV charging infrastructure market with a clear ambition: become Europe's leading fast-charging network by 2030.
What sets Electra apart is its dual identity: infrastructure company + tech company. Electra builds charging stations in the physical world. It's also a technology company that differentiates itself through the software powering both its charging points and mobile app. But bringing AI to an organization that spans physical infrastructure and software meant more than just picking the right tool.
The challenge: Overcoming internal resistance to find AI that works for everyone
When Electra first started exploring AI solutions, they ran into an unexpected obstacle. It wasn't budget or technical limitations. It was internal resistance from the people who understood AI tools best.
"At first, there was a preconceived idea on the tech side, saying that we don't need such kind of solution because other tools exist and are ‘easy to grasp’," recalls Marie, Electra's Chief People and Impact Officer who led the change management efforts for AI adoption.
The tech team's perspective made sense for them. They were already comfortable with various AI tools, knew how to prompt effectively, and could navigate different platforms without much friction.
But that view missed a critical reality about Electra's organization. "The tech team is one thing, and of course, they are much more exposed to AI," Marie explains. "And then you have all the rest of the organization, which is more than two-thirds."
This was about making AI actually usable for everyone, regardless of their technical background. The company needed an AI agent solution that could work for both engineers building charging infrastructure and for customer care agents answering invoice questions. For operations teams coordinating across 10 countries and for HR teams supporting career development conversations.
There was also another layer to the challenge. Electra's cybersecurity team needed visibility and control. Prior to Dust, Electra had distributed generic ChatGPT licenses across teams. But Cybersecurity needed "to centralize compliance, measure what data access was centralized and how users could access it," as Marie puts it. With scattered ChatGPT licenses, they had no way to track what information was being shared with AI tools or to enforce data governance policies.
The question facing Electra was clear: How do we make AI know the context of our company while ensuring it works for everyone and meets our security requirements?
Proving the concept with customer care
The answer came through Dust, a custom AI agent builder that offered native integrations with all their existing tools and a no-code interface that anyone could use.
Early in 2025, Electra had assembled a small interdisciplinary squad to chart the company's AI strategy. In April 2025, they launched a proof of concept with Dust. Customer care became the first use case.
In retrospect, if AI could help support the customer care team - who weren't expected to be technical experts - it would show that Dust could work for the entire organization.
The customer care team was facing a real scaling challenge. Electra was growing rapidly across Europe, and support ticket volume was increasing with it. The team had already automated responses for straightforward tickets, handling about half of the incoming volume. But the remaining tickets that required human attention were complex and time-consuming, requiring team members to pull information from multiple systems - Slack conversations, Notion documentation, and backend data about chargers and payments.
"We really needed to make sure that those complex tickets were solved the best way and the fastest way," explains Inès, who led Electra's AI implementation. "Also, that agents working in customer care were having their best added value in really resolving those complex use cases."
The solution: Three specialized customer care agents that passed the test
The team identified three categories of escalated tickets that consumed the most time from Electra’s customer care team: invoice-related questions, refund requests, and general complex inquiries about charging sessions or technical issues. Rather than building one generic support agent, they created three specialized Dust agents, each trained on the specific context needed for its domain.
Here's how the system works: When a human agent receives an escalated ticket from Intercom, they can call one of the three Dust agents depending on the ticket type. The agent automatically scans the conversation thread to understand the customer's issue. Then it pulls relevant information from three key sources:
- Slack conversations where teams discuss station issues, customer problems, and operational updates
- Notion knowledge base containing documentation, procedures, and troubleshooting guides
- Custom backend MCP (Model Context Protocol) that Electra built, which provides real-time access to charger data, charging sessions, payment methods, and customer account information
The key to specialization: Smart prompts, not separate data
All three agents access the same data sources. What makes each one specialized is how it's prompted:
- Refunds agent: Follows a decision policy with specific rules for different scenarios (promotions that didn't apply, payment type conditions, etc.)
- Invoices agent: Trained to investigate issues specific to Electra's invoicing module in their backend
- General inquiry agent: Uses a broader prompt to handle complex questions about charging sessions and technical issues
Within three minutes, the Dust agent delivers a pre-drafted response that includes all the necessary context, relevant links, and specific data points. The human agent reviews the response, makes any needed adjustments, and sends it to the customer.
Rolling out across the team: Champions and office hours
After validating the customer care use case during the POC phase, Electra rolled out Dust more broadly. They took a thoughtful bottoms up approach, starting with 50 licenses distributed to "AI champions" across different teams and locations.
This bottoms-up strategy was intentional.
"The goal really was to provide the means for people to experiment," Marie explains. "The bottoms-up approach was extremely important to ensure that we are not cutting initiatives by saying 'no, this is how we should do things' and 'this should be validated by a VP' whatsoever, because this is not how we believe in change.”
Marie and Inès wanted to create a virtuous cycle where early adopters would become evangelists within their teams. "We really spotted also that when you have an AI champion in a team, they also encourage the other team members to build with them," Inès notes.
For customer care specifically, having champions proved valuable because many of the agents worked externally or across different countries. The champions could provide local support and demonstrate the agents in action.
The team also launched office hours where any employee could get hands-on help with Dust. The office hours gave customer care agents dedicated time to experiment with the three support agents, ask questions, and share best practices with colleagues. This removed the "I don't have time to learn this" barrier and created a safe space for learning.
Results: 80% time savings on escalated tickets and higher quality responses
Since Dust’s launch, the three customer care agents have become the most heavily used agents across Electra. "I would say there's an average of 4 new conversations created per hour to one of those agents for customer care," Inès says.
The 80% time reduction per escalated ticket has had cascading effects. Customer care agents can now handle a higher volume of complex tickets without feeling overwhelmed. The time they save on information gathering and drafting lets them focus on the human elements of support: understanding customer frustration, making judgment calls on edge cases, and ensuring the response truly solves the problem.
The quality of responses has improved as well. Because the Dust agents pull from the latest information in Slack and Notion, agents no longer risk giving outdated information. And because the backend MCP provides real-time data about charging sessions and payments, responses are specific and accurate rather than generic.
Across the entire company, Electra has seen strong AI adoption. Within the first month of the September 2025 launch, they hit 70% weekly active users. That number has stabilized between 70-80% depending on seasonality. Since September, employees have created 170 different Dust agents for various use cases across the organization.
But for the customer care team, the impact is concrete and consistent. Escalated tickets that once took extensive searching and drafting now takes 3 minutes. The team has gained 80% time savings on answering complex tickets. This is truly a customer care operation that can scale with the business.
Scaling across Europe with confidence
As Electra continues its expansion across Europe, the customer care team now has the infrastructure to scale efficiently. They're handling more complexity, more languages, and more customers than ever before. And the 80% time savings on escalated tickets means human customer care agents can focus on what humans do best: solving complex problems and creating positive customer experiences.
"We wanted to make sure we have this customer care team that is able to grow with the number of volume tickets that is going to be more and more important," Inès reflects. With Dust, they've built exactly that.
Interested in learning more about how Dust can help your team? Visit our solutions page or reach out to our sales team.
