WorkEasy · AI & Workforce

Worky - AI Native Scheduler

An agentic scheduler for WorkEasy that builds a manager's shifts from their own rules, then hands back a draft they can question, edit, and approve. A vision piece designed to make AI scheduling something managers actually trust.

Role: Sole Product Designer
Timeline: 2 months · part-time
Year: 2025
Industry: AI · Workforce Management

The problem

Building a shift schedule by hand is one of the most draining parts of a manager's week. Availability, skills, overtime caps, and shifting demand all have to be held in one head at once, every week, and a single wrong call ripples straight into payroll and morale.

My role

I designed the concept end to end as the sole designer, guided by my manager, over two months of part-time work. It was built to pitch to the client, so the whole thing is a vision piece: a working design for how agentic scheduling could live inside WorkEasy.

The outcome

A scheduler that does the heavy lifting and hands the manager a draft to approve, not a black box to obey. Every AI decision arrives with a plain-language reason, so trust is designed in from the first screen.

90% less

Projected time to build a schedule, from hours to minutes

Scheduling jobs the agent handles, from forecasting to filling gaps

0 black boxes

Every AI decision shown with a reason the manager can read

Challenge

AI that managers will actually trust.

The promise of AI scheduling is easy to say and hard to earn. A manager will not hand their week to a system that produces a schedule with no explanation, because when it gets one shift wrong, cleaning it up is their problem. So the real challenge was never the automation. It was trust. The design had to let the agent do the heavy lifting while keeping the manager firmly in control, and it had to explain itself well enough that a manager would believe it before relying on it.

Process

Teach the agent, then keep the human in charge.

I designed the flow in two halves. First the manager teaches the agent the rules that matter to them: maximum hours and overtime caps, the skills or certifications each shift needs, who is available, and the coverage they have to hit. The agent draws on a knowledge base of past schedules, employee profiles, and historical demand to make its picks. Then it drafts. For the week or month the manager set, the agent produces a full list of shifts, choosing people based on their history, and hands it back. The manager stays the decision-maker at every step, free to edit a shift, reject it, or ask the agent to try again, and nothing is final until they say so.

Solution

A draft you can question, not an order you follow.

The finished concept is an agentic scheduler that feels like a capable assistant rather than an autopilot. It generates the full schedule from the manager's rules, forecasts how many people a period needs, fills gaps, and resolves time-off clashes on its own. Every suggestion carries its reasoning in plain words, so the manager sees exactly why the agent picked someone, reasons like available, qualified, and under the overtime cap. Around the scheduler sit the supporting AI moments I also designed: smart timesheet corrections, automatic time-off decisions, attendance anomaly detection, and shift-swap matching, each following the same rule, do the work, then show why. It is a vision for AI in workforce management that earns its place by being transparent, not only fast.