Your AI Agent Can Diagnose Your EV Charging Network: 3 Ways to Use eMabler MCP
November 16, 2026
Read time: 7 minutes
Author: Ville Parviainen, CTO at eMabler

Most teams answer questions about their charging network the same way: someone logs into a portal and starts digging.
Maybe they open the charger's page in the portal, scroll through its events, check the failed sessions and compare what they find with a ticket in another system. The answer is in there, but finding it takes time, and the person who asked is still waiting.
An AI agent could do that work for your team, but first it needs a connection to your charging data.
MCP is short for Model Context Protocol, the open standard AI agents use to connect to other systems. eMabler MCP is now live, and it gives your agent a read-only connection to your charging data. We built it read-only on purpose, so your agent can read whatever its key allows but cannot change anything.
That is why every way to use eMabler MCP starts with a question you ask your agent. The answers get more useful when the agent can also read your own systems, and it can, because it runs in your environment. That lets it pair your charging data with your tickets, your targets or your energy prices.
Of all the ways to do that, I picked the three examples to start with, and all of them work today.
Use case 1: ask what is wrong, and why
A driver calls, a site goes quiet or a charger starts misbehaving. In each case, your team can ask its own agent what is going on. Of the three ways, I would try this one first. Tracing a fault eats into your team's day, and here the agent does it for them.
The agent reads your charger’s live status, its recent events and fault codes, and the error history of the failed sessions. It then answers with the evidence attached. If it also potentially reads your ticketing system, it knows whether the charger already has a ticket open and what was tried last time.
Here are some of the questions eMabler MCP can already answer today:
- “What is the problem at site X right now?”
- “Why did this driver’s charging session fail this morning?”
- “Which chargers report faults, and since when?”
- “Has this charger failed the same way before?”
Each of those questions is about one fault at the moment it happens. Your agent can also answer a different kind of question: how all your chargers performed over a week, a month or a quarter.
Here is how that works.
Use case 2: run the fleet review without building reports
The numbers behind a healthy network are one question away. Your agent reads session success rates, utilization, consumption trends and summary metrics, per site or across the fleet.
They are the same statistics eMabler's portal shows, but your agent puts them next to your own targets in the tool your team already uses. As a result, the Monday review becomes a conversation with the data instead of a morning spent exporting it.
Here are some of the questions eMabler MCP can already answer today:
- “Which ten chargers had the worst session success rate this week?”
- “Which sites run close to full utilization, and when?”
- “How did consumption develop per site this quarter?”
- “Summarize fleet health for the ops meeting, compared to last month.”
Those questions cover how well your chargers performed. However, as the price of energy changes through the day, a session can succeed and still cost more than it should. Your agent can check that too, once it also sees your energy prices.
The third way to use eMabler MCP shows you exactly how.
Use case 3: put charging next to your energy prices
To check what a session cost, your agent needs two things. eMabler MCP gives it the first: how much energy your sites used, hour by hour. The second comes from your side, or potentially from eMabler tariff engine. Your agent puts the two together and works out what each hour of charging cost.
That means you can check your charging costs every day, without waiting for the invoice at the end of the month. You also see which sites charge during the most expensive hours, so you know where moving the load to cheaper hours would pay off.
Here are some of the questions eMabler MCP can already answer today:
- “Did yesterday’s charging run at good prices?”
- “Which sites charged through the most expensive hours this week?”
- “How much energy went through site X this month, hour by hour?”
Those questions cover what your charging costs. However, the people who run your chargers are not the only ones who need answers from your charging data. Your agent can answer other teams too.
The next section shows what support, finance and operations can ask.
More ways to use eMabler MCP in support, finance and operations
Faults, fleet performance and energy costs are three ways to use eMabler MCP, but they are far from the only ones. Other teams in your company depend on the same charging data and have questions of their own.
The nine examples below show what your other teams can ask their agent.
- Help the driver while they are still on the line. Support asks the agent for the driver’s chargers, their ongoing session and what the charger reported, together with the CRM record already in front of them. The answer arrives during the call instead of after a callback.
- Check the money before it moves. A finance agent reads invoices, billing periods and the sessions behind them, then compares them with your ERP before anything is sent or paid. A wrong price, a missing session or energy the meters never recorded is caught as a correction instead of a credit note.
- Feed the reports you already write. Usage trends, session statistics and cumulative cost go straight into the weekly management summary or board pack your agent drafts. Nobody has to fill in the charging slide by hand.
- Know what ran overnight. eMabler runs the automations your team configured, such as scheduled reboots, alerts and rules. Your agent reads those rules and their activity log. It can answer, “What did the automations do last night, and did anything need attention?” before the first coffee.
- Answer pricing questions on the spot. Your agent can read tariffs and their versions. Support can ask, “Which tariff applies at this site, and what changed in the latest version?” while the customer is still on the line.
- See the network live. Your agent sees the ongoing sessions across every site. “How many sessions are running, and which sites are busiest at this moment?” turns the morning status check into one question.
- Audit who can charge where. Your agent can list the authorized drivers per site and answer “Who has access to site X, and does that match our policy?” That is useful the week before an audit and the week after an offboarding.
- Keep your registers honest. Sites carry your own external IDs, so your agent can check “Does every site in our asset register exist in eMabler, and do the charger counts match?” Mismatches come back as a list instead of as surprises.
- Let the agent check its own limits. The agent can ask what its key is allowed to see. When it cannot answer a question, it can tell you that the key has not been granted that access.
Conclusion: what eMabler MCP does and how to connect your AI agent
Finding an answer in your charging data usually means that someone logs into a portal and searches for it. With eMabler MCP, your own AI agent does that search for your team, saving them plenty of time and manual work. The connection is read-only, so your agent can answer questions about your data but cannot change anything. We also use OAuth authentication tied to the Operator rights, so the agent can only see what the Operator sees.
There are three good places to start. The first is to ask what is wrong and why, and get an answer based on the charger’s status, events and fault codes. The second is to run the fleet review without building reports, with success rates and utilization next to your own targets. The third is to put your charging next to your energy prices and check your costs every day instead of at month end.
Other teams can use the same connection. For instance, Support can answer a driver during the call, Finance can check invoices before they are sent or paid, and Operations can see what ran overnight.
All of it needs only one setup. Add the eMabler MCP URL to your agent, sign in through eMabler Connect with your Operator account and it automatically creates the authentication key. The key never travels in a URL, you decide what it can see, and a revoked key stops working within a minute.
Reading is all an agent key allows today. Actions will come later, with some additional security gates to not let the agents go rogue on your charging fleet.
Connect your agent today and ask it one question your team asked this week. The eMabler MCP documentation walks you through the setup: docs.emabler.com/docs/emabler-mcp