A confession before the argument

I spent the better part of a decade in operations. Sky Regional first, then Etihad, then flyadeal during the launch period, then a film studio under Technicolor MPC. Different industries, the same job under the surface — get the right people to the right place at the right time, account for the ones who could not show up, and make the next shift work anyway.

For most of that decade I assumed our problem was the tools. We kept buying better ones. Better optimisers, better dashboards, better integrations with payroll. Every cycle the vendor would show us a chart of how much we had saved, and every cycle the duty manager would still be running the operation off a whiteboard.

I want to write down the thing I learned only after I stopped working inside the operation and started building tools for people who do. The tools were never the problem. The problem was the gap between the surface where we made the plan and the surface where the plan met the day.

What the planning surface actually does

A workforce planning system, at its core, is an optimiser. You feed it demand, you feed it constraints, and it returns a schedule. Some of them are very good. We had a tool at the airline that could solve a fleet-wide cabin-crew roster in minutes and respect every fatigue rule, every visa restriction, every pairing constraint, every per-base equity rule the union had negotiated.

None of that was the work.

The work began at 0500 the day the roster went live. By 0530, two crew had called sick and one had missed an inbound. By 0600 the duty manager had three problems to solve, none of which the optimiser was running on anymore, because the optimiser ran yesterday. The schedule the optimiser had produced was now a historical artefact. What the duty manager held was a snapshot of the plan, three exception lists in three other systems, and a phone.

The whiteboard is the truth

Every operation I worked in had a whiteboard. Sometimes literal, sometimes the laminated Day Sheet pinned next to the duty manager's monitor. The whiteboard was where the actual assignments lived after 0600. It was the only place where the plan as executed could be reconstructed.

Nobody was paid to write on the whiteboard. It was not in any workflow document. It did not exist in the system of record. And yet the whole operation depended on it, because the whiteboard was the only place that could keep up with the rate at which reality diverged from the plan.

At the studio it was a spreadsheet on a shared drive. At flyadeal it was a printed manifest with handwritten changes. At Etihad it was a Microsoft Teams chat between duty managers across three bases. At Sky Regional it was actually a whiteboard, with two markers — one black, one red — and the convention that red meant the change had not been entered into the system yet.

None of those were anomalies. They were the operating layer.

Why the tools never caught up

I have a theory about why every vendor in this market eventually ships the same product. They look at the planning surface, because that is the surface their buyer sees during the procurement. They watch a director of operations approve a roster in a meeting. They watch a workforce-management analyst tune a constraint. They do not watch the duty manager at 0600.

So the product gets better at the planning surface every release. New optimiser, new dashboard, new what-if module. None of it reaches the whiteboard, because the whiteboard runs at a different cadence and operates on different state.

The duty manager does not need a better roster. The duty manager needs a system that can absorb a change, propagate it to the seven places it matters, and tell the people whose lives just changed what to do next. The roster only matters insofar as it anchors that propagation.

The two surfaces

I now think of every workforce system as having two surfaces, and the gap between them is the entire job.

The first surface is the plan. It runs on a slow clock — weekly, monthly, sometimes a season at a time. The signals are forecasted demand, headcount, leave, training. The optimiser is the right kind of tool. The buyer is HR, finance, or operations at the director level. The audit question is whether the plan respects the policy.

The second surface is the operation. It runs on the clock of the day. The signals are sickness, no-shows, equipment unavailability, weather, a flight that diverted, an actor who cannot make the call sheet. The optimiser is not the right kind of tool, because you cannot rerun an optimiser at 0617 in the morning when the duty manager has thirty seconds to decide whether to delay the flight or pull from the standby pool. The buyer for this layer is the operator, and the operator is not in the procurement meeting. The audit question is whether the action was justifiable in the moment.

The vendors who win the procurement build for the first surface. The whiteboard is the second surface refusing to be optional.

What we are building, and why

The thing I wish we had when I was running operations is the substrate that sits underneath the second surface. Not a better roster. Not a better dashboard. A layer that knows the plan, watches the reality, owns the propagation when the two diverge, and writes down what it did so a duty manager at the end of the shift can hand off cleanly.

That is what MAIA is. The roster is one input. The fleet tracking system is another. The crew bidding tool is a third. The phone the standby crew member is about to answer is a fourth. The substrate fuses those into one operating picture, drafts the action the operator was about to take anyway, surfaces the proof, and stays out of the way once the operator approves.

The whiteboard becomes the system of record, not an off-system workaround. The operator stops being the integration layer between five tools. The audit chain gets a record of the action as taken, with its reasons, in a form an inspector can replay.

What I would tell my younger self

If you are the head of workforce operations and you are looking at a procurement deck right now, the question I wish someone had taught me to ask is this. Show me what the duty manager does at 0600 on a Tuesday when the roster is already wrong. Walk me through the screens. Show me where the changes go. Show me where the whiteboard would have been, and what replaces it.

Most vendors cannot answer that question, because they have not built for that surface. The ones who can are the ones who understand that the schedule is the input, not the deliverable. The deliverable is the next thirty minutes.

That is the part nobody writes down. I am writing it down now.