Reliability intelligence for EV charging networks

Broken chargers don't always look broken.

Chargers that say available but deliver nothing. Sessions that never start. Energy quietly leaking out of your network. Tadit finds these faults, files the issue where your team works, and explains the why when you ask. All on top of the CPMS you already run.

SaaS or private deployment · every credential encrypted with keys unique to your workspace

13
fault types
caught out of the box, nothing to configure
kWh
every fault costed
energy lost, per station, per cause
Minutes
to root cause
the AI agent investigates, not your engineers
24/7
playbooks act
fault to ticket, instantly · no one on call
MCP
ask your network anything
Claude, ChatGPT or Gemini query your fleet, read-only
Your
CPMS
OCPP 1.6 & 2.0.1 · no new hardware, no migration

Every charging network bleeds quietly.

A driver pulls up, plugs in, gets nothing, and leaves. Your dashboard stays green. No alarm. No ticket. Just lost revenue, and a driver who won't come back. When someone finally notices, an engineer loses half a day in the logs. Most of these faults are never investigated at all.

You can't fix what you can't see.

How it works

Tadit is reliability intelligence on top of your CPMS.

It's not another charging platform. It watches the one you already have, and does three things, continuously, for every charger you operate.

1

It sees every fault. Even the silent ones.

Thirteen fault types caught out of the box, plus the ones no rule can catch: every charger learns its own normal, so drift, a firmware version failing as a group, or sessions ending the wrong way get flagged early. Faults are ranked by how busy the charger is, so you fix what drivers feel first.

Fault queue · busiest chargers first
Charger 04
delivered nothing · 9 sessions this week
busy charger
DC180 fleet
firmware failing as a group
12 chargers
Depot North
gone quiet · usage below its normal
flagged Tue

Every fault: 45-day history · event log attached

2

It finds the why in minutes.

Ask the AI agent about any flagged charger, and it investigates the way your best engineer would: reads its history, checks the manuals, rules things out, and shows its work. Half a day of log-digging becomes a five-minute read.

Investigation · Charger 04
Why is charger 04 delivering nothing?

The meter stopped recording after last week's firmware update. Sessions run, but no energy is counted. The vendor's bulletin recommends rolling back to the previous version.

Confidence: high Every step shown
3

It files the issue where your team works.

Every fault arrives as one well-briefed ticket in Jira, ServiceNow or Slack, or a webhook to any system you run, with its history and severity attached. Not a hundred raw alerts for your team to wade through.

Issue · filed automatically

Charger 04 delivers no energy · 9 sessions, 0.0 kWh

Severity high · worsening · 4th recurrence
Attached 14-day history and event timeline

→ Jira→ ServiceNow→ Slack→ Webhook
The diagnostic agent

Half a day of engineering. Replayed in minutes.

Point the agent at any charger, a fresh flag or a question of your own, and it investigates like your best engineer on their best day: reads the charger's history, weighs the likely causes, checks your manuals and service bulletins, and lands on a verdict in minutes. You decide when it runs.

  • It shows its work. Every hypothesis, every check, every conclusion is logged. The verdict comes with its confidence, and a trail you can defend in a vendor or warranty dispute.
  • It gets sharper with every fix. Confirmed verdicts feed back into detection. Reliability compounds.
Investigation · Charger 04 · live
Why is charger 04 delivering nothing?
Checked 14 days of history: 9 sessions, zero energy
Two suspects: stuck contactor, frozen meter
Meter unchanged while charging: contactor ruled out
Found it in your docs: vendor bulletin SB-112
Meter frozen after firmware update. Confidence: high. Recommended fix: roll back firmware.

18 steps · every one logged and reviewable

Ask your fleet over MCP

Ask your network anything. In plain language.

Tadit is an MCP server. Connect the AI assistant your team already runs, Claude, ChatGPT or Gemini, and ask straight away. It answers from live fleet data, read-only and scoped to your workspace. No dashboard to learn, no query to write.

  • Your assistant, your keys. Claude, ChatGPT or Gemini connect over MCP. Nothing leaves your workspace, and access stays read-only by design.
  • Answers you can trust. Secured with OAuth 2.1 and scoped per workspace. Your team asks; the fleet answers with numbers, not guesses.
Ask your fleet · live

via MCP · OAuth 2.1 · read-only · scoped to your workspace

Playbooks

Set it once. It runs every time.

Decide once what should happen for each kind of fault. Tadit does it every time after that: instantly, at 2 a.m., without anyone on call.

  • Any fault, any action. Tickets in Jira or ServiceNow, alerts in Slack, or a webhook that tells your CPMS to reboot the charger.
  • One ticket, not a ticket storm. A chronic fault becomes a single well-formed ticket with its history attached, not a hundred duplicates.
  • No code, fully logged. Your ops team sets the rules in the dashboard; every action is recorded and reviewable.
Playbooks · active
Delivers nothingJira ticket
Goes offlineSlack #field-ops
Hardware faultServiceNow incident
Repeats 3×Reboot via your CPMS

Deduped · one ticket per fault, history attached

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

Tadit measures the energy each fault leaks, per station and per cause, so you fix the most expensive problems first and walk into your ops review with numbers, not anecdotes.

What faults cost · this month4,116 kWh lost
Airport Eastdelivered nothing
1,284 kWh
Hauptbahnhofoffline
947 kWh
Depot Northstuck sessions
713 kWh
Mall Eastmeter frozen
612 kWh
Highway A1hardware fault
560 kWh

Works with any CPMS · vendor-neutral

Live in days. Not quarters.

Tadit connects to your charging platform, or straight to your chargers. No new hardware, nothing replaced. First alerts the same day, while behavioural baselines build in the background. New platforms are added on request, and your reliability data stays vendor-neutral, so you can compare hardware on facts.

Any CPMSDirect to chargersYour platform · ask us

Deploy it your way.

Use the hosted SaaS and be live in days, or run Tadit as a private deployment inside your own infrastructure when your data can't leave home. Same product, same detection, either way.

Your data stays yours.

Every credential your workspace holds, CPMS tokens, AI keys, integration secrets, is encrypted with keys unique to your workspace. No shared secrets, no cross-workspace access.

For your IT team: OCPP 1.6 & 2.0.1 · MCP server (OAuth 2.1, read-only) · Enterprise SSO · SaaS or private deployment · Encrypted at rest, keys unique to your workspace · Your own AI keys (OpenAI or Gemini)

Every incident makes the next one cheaper.

Tadit spots the fault, files the issue, and explains the why when you ask. Every confirmed verdict makes it sharper. That's the loop.

Book a demo

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