The AI-powered operations platform for EV charging networks

Your chargers show green. Some of them stopped earning weeks ago.

Tadit turns the OCPP stream your chargers already send into early warnings, real-time tickets and root causes. The result: more uptime, more energy delivered, fewer site visits.

Three numbers you already report. Silent faults hurt all of them.

Availability

A charger that answers heartbeats but delivers nothing still counts as available. Your uptime figure stays high while the site earns nothing.

Cost of reactive service

A visit that finds nothing wrong, then a second visit two days later, is two truck rolls for one fault nobody understood.

Revenue per charger

Drivers learn which chargers fail and stop coming back. The charger is still online. The revenue is not.

Millions of messages. One list.

A thousand chargers send millions of OCPP messages a day. Your platform shows the last one: online or offline. Tadit reads all of them and hands your team a short list, worst first.

Watch the shape, not just the snapshot.

1
Detect

Every fault in real time, with the charger's own error code. Then the quiet ones: a charger drifting from its own normal, a model failing as a group, a connector drivers are avoiding.

2
Explain

An AI investigator does the root-cause work an engineer would and ends with a confidence level. Every step is logged and reviewable.

3
Act

One ticket per fault in Jira, ServiceNow or Slack, with the history attached. Cooldowns stop re-ticketing. A network-wide outage holds tickets and emails the on-call.

millions in · five out
Today's list · ranked by how busy the charger isexample
1Charger 04 · Airport Eastdelivered nothing across 9 sessions · busy charger
2DC180 · firmware 3.2.1failing as a group · 12 of 40 chargers
3Depot North · plug 2usage fell below its own normal
4Hauptbahnhof 07returning fault · third time this month
5Whole fleetabnormal stops rising in the session mix
The daily reality

Three things that happen every week. What changes with Tadit.

1

The charger that was green all week and delivered nothing.

Drivers plug in, nothing flows, they leave. Nobody calls. At month end the site earned half what it should have.

With TaditZero-energy sessions are flagged, the kWh that should have flowed is counted, and the ticket is ranked above quieter chargers.
2

The Monday morning alert storm.

Two hundred alerts overnight. Most are three chargers repeating the same fault every fifteen minutes. The real problem is on page four.

With TaditRepeated events become one fault, one ticket. The same fault never re-tickets, and tickets are held during a network-wide outage.
3

The site visit that found nothing wrong.

A technician drove ninety minutes. The fault cleared itself an hour before. Two days later it was back.

With TaditA fault that comes back is marked returning, not new. Return often enough and it is a repeat offender: replace, don't reset.
Predictive maintenance powered by AI analytics

Sees the slide before the fall. Shows you the evidence.

Every learned signal moves while the charger is still online and still listed as available. That is when a visit gets planned instead of dispatched.

  • A charger drifting above its own normal. Degrading. Schedule the visit before it goes dark.
  • A make, model and firmware group shifting together. A batch problem. The cohort alarm lists the units that have not failed yet.
  • A question, answered with a trail. The investigator forms hypotheses, checks history and your docs, and concludes with a confidence level. Every step logged, defensible in a warranty conversation.
Learned signalsexample data
JulAugSep
Normal rangeFaults a day, this chargerShift started
Baseline0.4 faults a day
Shift5× its normal
Started12 Aug, still online

Know what every fault costs. To the kilowatt-hour.

Tadit turns fault time and failed sessions into energy that did not flow, per charger, per fault type, per day, and ranks chargers worst first. So you fix the most expensive problems first and walk into your ops review with numbers, not anecdotes.

It states kWh. Your tariff is yours, so the conversion to money is yours too.

Energy lost · last 30 days example4,116 kWh
Airport Eastzero-energy sessions
1,284 kWh
Hauptbahnhofoffline
947 kWh
Depot Northabnormal stops
713 kWh
Mall Easthardware fault
612 kWh
Highway A1component fault
560 kWh
MCP server · AI-powered operations

Ask your fleet anything. In plain language.

Tadit is an MCP server. Add it to Claude, ChatGPT or Gemini as a read-only connector, sign in with your company login, and ask. The answer comes from live fleet data in one sentence, so nobody opens three tools and promises to get back to the boss.

  • Twelve read-only tools. Fleet summary, one fault in full, energy lost, session outcomes over 7, 14 or 30 days, connectors with unusual occupancy, cohorts with alarms, and any charger by its sticker code.
  • Your access rules apply. OAuth 2.1, with a token valid only for this surface. Revoke a user and the connector is revoked with them.
  • A second MCP surface for the driver on the phone. An AI voice line takes the sticker code the driver reads out, sees the live state of that exact connector, and walks them through the safe steps first. With the driver's spoken consent it can stop a stuck session or release the cable, through your management platform. Every action is logged. It refuses to reboot while another plug is charging.
Ask your fleet · over MCPexample

MCP · read-only · OAuth 2.1 · your company login

OCPP-native. Vendor and hardware neutral.

Nothing to install. Nothing replaced.

Tadit reads the raw OCPP message stream your chargers already send: boot notifications, heartbeats, status, transactions, meter values, firmware and network status. OCPP 1.6J and 2.0.1, including OCMF signed meter readings. Any charger make, any CPMS. The feed is one-way into Tadit. The only outbound actions are the ticket integrations you configure.

OCPP 1.6JOCPP 2.0.1OCMF signed readingsAny CPMSAny charger makeNo hardware at the site

Faults in real time. Your first report in four weeks.

Every fault is detected in real time from the first message. The learned signals, each charger's own normal, its cohort and the fleet's session mix, need 28 days of history to be fully weighted. Chargers with peers already in the workspace are useful sooner.

1
Day 1
Connect

Point your platform's OCPP event stream at Tadit. Each customer gets an isolated workspace and their own API key.

2
Day 1
Register chargers

Each charger is registered with its sticker code, the id drivers and support already use.

3
Days 1 to 28
Learn

Faults are flagged in real time from the first message. Baselines build in the background and are fully weighted at 28 days.

4
Day 28
First report

A ranked list of chargers losing energy, the repeat offenders, and any cohort behaving worse than its peers.

5
Then
Act

Connect playbooks to your ticketing tool and connect your AI assistant.

Start with a read-only pilot. Your fleet's first energy-loss and repeat-offender report in four weeks, on the chargers you already run. Fault tickets from day one.

Security and deployment.

Hosted, or in your own infrastructure.

Use the hosted service and be live in days. Tadit can also be deployed in your own infrastructure on request, when your data cannot leave home. Same product, same detection, either way.

Built for many operators. Isolated for each.

Every record and every job is scoped to your workspace. Credentials for ticketing tools are encrypted at rest. Single sign-on with your company login, Google and Microsoft included. User sessions, assistant connectors and the voice line each have their own credentials, and a token from one is rejected by the others.

Questions operations teams ask first.

Plain answers, no jargon.

How does Tadit know what normal is for my charger?

It learns from the charger's own recent history: how often it faults, how its sessions end, how busy it is. While that history is short, under 28 days, the baseline borrows from chargers of the same make, model and firmware, so a new charger is never blind.

Is this predictive maintenance?

Yes. Predictive maintenance powered by AI analytics, in the operational sense. Tadit sees a charger's fault rate shift, a model behave worse than its peers, or a fault keep returning, while the charger is still online. That is when maintenance gets planned instead of reactive. Tadit does not quote a number of days before failure, because it will not publish a lead time it has not measured at fleet scale.

Will it reboot my chargers?

Only through your management platform, only from the driver phone line, only with the driver's spoken consent, and never while another plug is charging. The monitoring feed itself is one-way into Tadit.

Which platforms does it work with?

Any CPMS. Tadit reads the OCPP event stream your platform already carries, so the platform brand does not matter. Connecting yours is a small piece of integration work on our side, not a rebuild.

Where does my data live?

In the hosted service, or in your own infrastructure on request. Either way, every record is scoped to your workspace.

Statistics find it. AI explains it. You see both.

Start with a read-only pilot. Faults are flagged in real time from day one. Four weeks in, you have your fleet's first energy-loss and repeat-offender report, with every flag showing the baseline it broke from.

30 minutes. We bring the product, you bring the questions.