Proactive AI Operations

The Problem

Why does this extension exist?

Most operations teams find out about a problem when a driver complains. The charger that has been failing one session in four for a fortnight, the token being used in two cities on the same afternoon, the site whose payment failures crept up after a firmware update — none of these are visible on a dashboard unless somebody already suspects them. By the time they surface, the revenue and the goodwill are gone.

The Solution

What you get

Proactive anomaly detection across chargers, sessions and payments, with acknowledge, dismiss and resolve
A daily digest by email listing the critical issues and the actions worth taking
Root-cause analysis of failed sessions, instead of a list of error codes
Fraud detection — token cloning, free-charging and payment fraud — with an investigation and resolution trail
Network expansion recommendations drawn from where demand is actually being turned away
A network health score from 0 to 100, with history and trend
Seven assisted operational actions, each with a risk level, a required role, and explicit approval before anything but a status check runs

🎯 Who is this for?

Operators whose network has outgrown the dashboard, teams running lean out of hours, and anyone who would rather be told about a failing charger than told about it by a driver.

Enable This Extension Browse All Addons

🔌 Modular by design

Enable or disable per tenant. Pay only for what you use. All addons integrate seamlessly.

Key Capabilities

Everything included

Anomalies Found for You

Chargers, sessions and payments are watched continuously. What comes back is a short list worth reading, each item with acknowledge, dismiss and resolve — not an alert storm.

A Daily Digest

One email each morning with the critical issues and the recommended actions. Recipients and send time are yours to set; the reading takes two minutes.

Root-Cause Analysis

Failed sessions are analysed for what actually caused them, so the answer is 'this connector's lock fails when it is cold' rather than a column of error codes.

Fraud Detection

Token cloning, free-charging and payment fraud are flagged as alerts you can investigate and resolve, with the trail kept. Thresholds are yours to set.

Health Score and Trend

A single 0-to-100 figure for the network, with its history. It is the number to put in front of a board, and the trend tells you whether last quarter's work paid off.

Actions That Ask First

Status check, soft reset, hard reset, unlock connector, open a work order, take a charger out of service, change a location's tariff. Every one but the status check waits for your explicit approval.

Use Cases

See it in action

How operators around the world can use this extension to solve real problems and grow their business.

The charger nobody had reported yet

A unit starts failing one session in four after a firmware change. Anomaly detection raises it on day two, root-cause analysis points at the connector lock, and a work order is opened — a fortnight before the first driver would have complained.

A token in two cities at once

The same token authorises a session in Porto and, forty minutes later, in Braga. Fraud detection raises a cloning alert with both sessions attached, and the investigation closes with the token blocked and the CDRs disputed.

A restart that waited for a human

The assistant proposes a hard reset on a charger stuck in a faulted state. The action carries its risk level and waits. The duty engineer, who has the role for it, approves — and the reset runs with the approval recorded against it.

Simulated Case Study

Operator catches 71% of charger faults before a driver reports them

A network of 300 chargers learned about most failures from support tickets. After enabling proactive AI operations, anomaly detection and the daily digest moved detection ahead of the complaint: 71% of faults were raised internally first, and mean time to acknowledge fell from 19 hours to under 2. Fraud detection also surfaced a cloned-token pattern that had been running for three months against a single site.

71%

Of faults found before a complaint

< 2 h

Mean time to acknowledge

Assisted actions, all approval-gated

0-100

Network health score