GSked AI Agents
by GardaWorld
AI agents that fill open security shifts automatically — matching the right guard to the right post before a gap becomes a liability.
TL;DR
- Designed the interaction model for AI agents that propose, fill, and rebalance guard schedules.
- Turned a manual, spreadsheet-driven scramble into a supervised, explainable automation.
- Role
- Designer
- Platform
- Web
- Tools
- Figma
Scheduling thousands of security shifts across sites, certifications, and availability is a daily firefight. GSked introduces AI agents that handle the routine — surfacing coverage gaps and proposing the best-qualified guard to fill them. My focus was the trust layer: how a scheduler stays in control of, and understands, what the AI is doing.
Designing automation people are willing to trust
- 01
Discover
Interviewed schedulers drowning in last-minute call-offs. The fear wasn't AI taking the job — it was AI making an unexplained decision they'd have to defend.
- 02
Define
Set the principle: agents propose, humans approve, and every suggestion is explainable. Designed for supervised autonomy, not a black box.
- 03
Design
Created an agent activity feed, a 'why this guard' rationale panel, and clear confidence and conflict states. Made overrides effortless so control never left the scheduler.
- 04
Ship
Rolled out with a shadow mode where agents suggested but didn't act, building trust before enabling auto-fill.
Coverage gaps closed faster, with humans in the loop
−70%
time to fill a shift
100%
decisions explained
1-tap
scheduler override
Schedulers shifted from manually chasing coverage to reviewing and approving agent proposals — closing gaps faster while keeping accountability.
