The AI-powered operations platform for EV charging networks
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.
Flagged it on Tuesday, filed the ticket, and ranked it above quieter chargers. One question to the investigator found the frozen meter.
A charger that answers heartbeats but delivers nothing still counts as available. Your uptime figure stays high while the site earns nothing.
A visit that finds nothing wrong, then a second visit two days later, is two truck rolls for one fault nobody understood.
Drivers learn which chargers fail and stop coming back. The charger is still online. The revenue is not.
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.
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.
An AI investigator does the root-cause work an engineer would and ends with a confidence level. Every step is logged and reviewable.
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.
Drivers plug in, nothing flows, they leave. Nobody calls. At month end the site earned half what it should have.
Two hundred alerts overnight. Most are three chargers repeating the same fault every fifteen minutes. The real problem is on page four.
A technician drove ninety minutes. The fault cleared itself an hour before. Two days later it was back.
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.
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.
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.
MCP · read-only · OAuth 2.1 · your company login
OCPP-native. Vendor and hardware neutral.
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.
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.
Point your platform's OCPP event stream at Tadit. Each customer gets an isolated workspace and their own API key.
Each charger is registered with its sticker code, the id drivers and support already use.
Faults are flagged in real time from the first message. Baselines build in the background and are fully weighted at 28 days.
A ranked list of chargers losing energy, the repeat offenders, and any cohort behaving worse than its peers.
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.
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.
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.
Plain answers, no jargon.
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.
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.
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.
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.
In the hosted service, or in your own infrastructure on request. Either way, every record is scoped to your workspace.
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.