How It Works

From onboarding to payroll — six steps

One connected flow across the employee journey. No hardware at the gate, no shared scanner queue, no disconnected tools — and AI that prepares while people decide.

Step 01
Set up your sites
Draw a geofence around every site, post, ward, or branch. Set its shift pattern and working-day calendar. Assign a responsible manager. Most organisations finish this in under a day; a multi-site contractor can clone a template for each new site.
- Per-site geofence radius
- Per-site shift pattern + rest rules
- Per-site manager assignment
Step 02
Onboard your staff
Bulk import from Excel, photograph each staff member in good light for face enrolment, and capture document expiry dates (visa, labour card, medical, professional licence). Consent for biometric processing is captured at this step and timestamped.
- Excel bulk import
- Photo enrolment in 60s per person
- PDPL consent timestamped
Step 03
Daily check-in
Staff open the app at the work location, take a live selfie, and confirm. GPS + face match run together in seconds. The live operational dashboard updates in real time for supervisors; out-of-zone or low-confidence check-ins are flagged.
- Works on any modern phone
- Offline-capable — syncs when back online
- Live dashboard for supervisors
Step 04
Review exceptions
Anomalies — sudden swings in check-in time, unexpected device, repeated out-of-zone — surface to managers, not into silence. Smart suggestions speed routine decisions; every consequential call still needs a manager's click.
- Anomaly detection, not silent acceptance
- Smart suggestions, human approval
- Audit trail of every decision
Step 05
Export to payroll
At month-end, working hours, overtime, leave, and absences export in a format your payroll system or bank WPS gateway accepts cleanly. AiyaHR does not process payroll directly — it feeds it.
- WPS-ready hour exports
- Configurable overtime rules
- Inspection-ready audit log
Step 06
Report & improve
Operations leaders see trends across sites: which posts run hot, which shifts produce overtime, where document expiries cluster. The same data the supervisor uses daily is what the board reviews quarterly.
- Trend reports per site and per shift
- Coverage analytics for planning
- Document-expiry forecasting

What rollout actually looks like

The six steps above describe the system running. Getting there is a different question, and it is where deployments succeed or quietly fail.

Start with one site, not all of them

A single site for two weeks surfaces every assumption that was wrong — the geofence that is too tight, the shift pattern nobody mentioned, the supervisor who needs an approval right. Fixing those once beats fixing them everywhere.

Enrol faces before go-live, not on the morning

Face enrolment takes a couple of minutes per person and cannot be done at 6am while a shift is starting. Doing it in advance is the single largest predictor of a smooth first day.

Tell staff what is collected and why

Consent obtained without explanation is compliance, not consent, and it produces resistance that gets blamed on the software. Five minutes explaining retention and what is not tracked prevents most of it.

Run parallel for one pay cycle

Keep the existing process alongside for a single cycle and compare the outputs. Discrepancies found by comparison are cheap; the same discrepancies found by an employee querying their pay are not.

Give supervisors the exception queue on day one

Exceptions are the system asking a human a question. If nobody owns that queue in week one, they accumulate into a backlog that arrives as payroll queries at exactly the wrong moment.

Roll out the remaining sites in batches

Batches rather than all-at-once, so a problem at one site is contained rather than becoming a problem everywhere simultaneously.

None of this is technically difficult. Deployments stall on availability rather than technology — a supervisor who cannot spare an hour, a decision nobody is authorised to make. Knowing that in advance is most of the fix.

Common questions about the flow

What happens when there is no signal on site?

Check-ins cache on the phone and sync when connectivity returns. The timestamp recorded is the check-in time, not the sync time, which matters on remote sites where the two can be hours apart.

What if someone forgets their phone?

A supervisor records the check-in on their behalf, and it is flagged as a manual entry rather than passed off as a verified one. The distinction is the point — an unmarked manual entry is how attendance data stops being evidence.

Do staff need to keep the app open?

No. It is opened to check in and closed again. There is no background tracking through the working day, which is a deliberate design decision rather than a missing feature.

How long does a check-in take?

Seconds. Open, confirm location, take the selfie. Anything longer than that at a shift change with forty people queuing is a system that will be worked around.

What counts as an exception?

Out-of-zone check-ins, unusual times, a device that does not match the usual one, or a face match below the confidence threshold. Each surfaces for a human rather than being silently accepted or silently rejected.