Patrol logs and observations
Voice-captured at the point of patrol, structured, timestamped and geolocated — rather than written up at the end of shift from memory.
AI agents for security and guarding services
In one line: security AI is emerging rather than mature. The use cases are real and the ROI on triage is strong, but the vendor landscape is fragmented and the cost of a false negative is high — so human-in-the-loop is the standard deployment pattern, and we build to it.
Last updated:
The administrative reality
Guarding is a margin business built on rosters, licensing and reporting. The operational work is the guard’s; the profit is decided by whether shifts are covered compliantly, incidents are documented defensibly, and clients receive reporting that justifies the contract.
All three are currently manual, and all three are agent work.
Workflows we automate
Incident reporting
Guided capture producing a defensible report, with severity triaged and escalation immediate where warranted.
Rostering against licensing
Shift coverage checked against licence currency, site-specific requirements and fatigue rules as hard constraints.
Client reporting
Contract reporting assembled from captured activity rather than compiled manually each month.
Licence and training currency
Tracked continuously across the workforce, flagged well before it blocks a roster.
Traditional vs agent-augmented
| Today | With an agent | |
|---|---|---|
| Patrol log | Paper or app, end of shift | Voice at the point, geolocated |
| Incident report | Free text, variable quality | Guided, structured, defensible |
| Shift coverage | Manual against a spreadsheet | Licence-checked, constraint-enforced |
| Client report | Compiled monthly | Assembled continuously |
| Licence expiry | Found when it blocks a shift | Flagged weeks ahead |
What we integrate with
- Workforce management: Deputy, Humanity, industry rostering platforms
- Guard tour and patrol systems: existing patrol platforms
- Incident and case management: where deployed
- Capture: ruggedised mobile, voice, offline-tolerant
How we de-risk it
- Human-in-the-loop is the standard pattern here, not an option. The cost of a false negative on a real threat sets a higher bar for autonomy than most sectors.
- No autonomous threat assessment. Agents capture, structure and route. Guards and controllers assess.
- Licensing constraints are absolute. Never traded against coverage.
- Incident records are immutable and versioned, because they end up in court.
Related use cases
The same workflows, described by the job rather than the sector.
Frequently asked questions
Does this do threat detection?
No. That’s a different product category with a different risk profile. We build the administrative and coordination layer.
Will incident reports hold up legally?
They’re stronger than free-text reports written hours later — timestamped, geolocated, structured, with provenance on every element. Your legal team should review the format before deployment.
Can guards use voice capture on site?
It’s tested in shadow mode with your actual conditions. Where a site makes voice impractical, we use guided structured capture instead.
What about privacy?
Records concern activity and incidents, not staff surveillance. That distinction has to be real, and how you introduce it matters as much as the build.
Not sure this is your operation
Every sector page describes the workflows we’d expect to find. Yours will differ in ways that matter, and the audit is designed for exactly that — you get evidence about your own business rather than someone else’s.