RESOURCESReference· How it works
The mechanics, written down.
A fleet platform asks you to trust it with fuel money and with the record of where your vehicles went. That is easier to do when the mechanism is legible, so here it is: the checks, the failure modes, the interfaces and the vocabulary.
01How it works· Mechanism
Four things worth understanding before you rely on them.
When the network drops
Connectivity here is not a reliability metric, it is a condition. Every surface has a defined behaviour for losing it, and the reconciliation always trusts the clock nearest the event.
Stores positions on the device — tens of thousands of records, depending on the unit — and forwards them when coverage returns. Timestamps come from the device clock, and duplicates are rejected on device and time, so a replayed buffer cannot inflate a trip.
A buffered gap is evaluated retrospectively when it lands. An overspeed that happened out of coverage is still an overspeed, raised with the time it occurred.
Authorizes against locally cached card rules, stores the sale, and syncs on reconnect flagged for review. A forecourt does not stop selling because a tower is down.
Trips, work orders, inspections, incidents and DUI tests are captured and queued. Biometric re-authentication stays valid offline for 24 hours.
Vector tiles for Liberia, Sierra Leone and Guinea are pre-downloaded, so the map still draws with no data at all.
Driver safety scoring
A score is only useful if the driver can see how it was arrived at and do something about it. Events are weighted rather than counted flat — a failed breathalyser test is not the same as one harsh corner — and the driver sees their own score, their own events and the trend in their app, not only their manager.
Measured
Speeding, harsh braking, harsh acceleration, harsh cornering, idling and fatigue, normalised against distance driven so a busy driver is not punished for being busy.
Acted on
An event can be dismissed or turned into coaching, which assigns training matched to the weakness and tracks whether it was acknowledged and completed.
Contestable
Penalties tied to events carry the underlying evidence, with timestamps and positions, so a deduction can be argued with instead of simply appearing on a payslip.
Remote immobilisation
Cutting a vehicle’s ignition remotely is the most dangerous thing this platform can do, so it is the most heavily gated. The command refuses to run above 5 km/h, refuses to run on telemetry older than five minutes, requires a typed reason, and writes the speed and position at the moment it was issued into the audit log alongside who issued it.
Those two gates are not configurable. A driver is never immobilised by surprise at speed, and the analytics layer can recommend but never act — a human with the permission has to issue it.
02Build on it· Interfaces
An API, signed webhooks, and rules you can write without us.
Developer API
https://api.fleet.kappagee.com/v1/ — one domain. Fuel lives under /v1/fuel, not on a second API.
A scoped API key, sent as a bearer token. Keys are shown once at creation, revocable, and rate-limited per account.
Every scope maps to real endpoints — vehicles:read, trips:read, alerts:read, fuel:read. A key without the scope gets 403, not a partial answer.
Outbound events signed with HMAC-SHA256 so you can verify the sender. Delivery outcomes are recorded rather than dropped.
SCIM 2.0 for users from your identity provider, and OIDC single sign-on. Deactivating someone there deactivates them here.
Automation, without writing code
Rules are built as an event, a set of conditions and one or more actions — notify, raise, require an approval, or call a webhook. They can match on all conditions or any of them, run in a priority order, and hold a cooldown so one noisy vehicle cannot fire the same rule fifty times in an hour.
You can see whether it fired
Every match writes a run — the rule, the event, the subject, what it did and whether that succeeded. A rule that has never fired says so on its own row. An automation you cannot audit is not an automation, it is a rumour.
You can test it before enabling it
A dry run evaluates the conditions against a sample and shows which passed and whether the rule would have fired — with nothing executed and nothing written.
Outbound webhooks are restricted to public addresses and signed, so a rule cannot be pointed at something inside our own network and used to read it.
03Reference· Vocabulary
The tables people ask us for.
Report catalogue
Twenty-five standard reports across fleet, safety, fuel, finance and compliance. Each runs live against your own data over a period you choose, and downloads as CSV, Excel or PDF. Anything not in the list can be built in the no-code builder and scheduled like the rest.
- Fuel usage by vehicle
- Trip log summary
- Maintenance cost
- Driver violations
- Device uptime
- Work-order SLA
- Driver safety scorecard
- Fleet utilisation
- Geofence event log
- Penalty summary
- Document expiry status
- Vehicle inventory
- Driver roster
- Trip activity summary
A representative selection · the full catalogue is in the platform
Roles & permissions
Roles are enforced in the API guard, not merely reflected in the interface — hiding a button is not access control. Custom roles can be composed from the same permissions.
| Role | Sees | Can do |
|---|---|---|
| tenant_admin | Whole account | Users, settings, billing and every module. |
| fleet_manager | Whole account | Vehicles, drivers, trips, maintenance and fuel policy. |
| regional_manager | Their region | A fleet manager, scoped to an assigned part of the fleet. |
| dispatcher | Whole account | Live tracking, trips and work orders. No administration. |
| mechanic | Whole account | Maintenance only. |
| accounts_officer | Whole account | Fuel spend, penalties, settlement and payroll exports. |
| auditor | Whole account | Read-only. The guard refuses unsafe verbs outright, not just the buttons. |
| driver | Their own record | Mobile app: own trips, score, fuel card and profile. |
| station_partner | Their station | Partner portal: their own forecourt and nothing else. |
Vehicle status vocabulary
Seven states, one vocabulary. The same seven drive the map markers, the sidebar filters, the wallboard and every status count in reporting — so two screens never disagree about what a vehicle is doing.
NEXTTry it against your own fleet
Reading about it only gets you so far.
We will set up your vehicles, put trackers on them and let you watch the record build itself for a week. If it does not tell you something you did not already know, that is a fair answer too.