Dynamic scheduling for facilities maintenance: what AI changes
Most maintenance teams still plan the day once, in the morning, and then spend the rest of it re-planning by phone. Dynamic scheduling is the alternative: a plan that changes as the day does. This guide explains what it is, why it was impractical until recently, and how to judge a dynamic scheduling tool.
Published
The short answer
Dynamic scheduling assigns and re-plans engineers' work continuously as conditions change: an urgent job comes in, a visit overruns, a customer cancels, an engineer is off sick. A fixed daily plan cannot absorb that, so the day falls back to whoever is free and nearest. AI makes dynamic scheduling practical at scale, because it can weigh the whole workforce against every job and keep doing so as the day moves. TYTEN offers this as AI engineer scheduling, new and ahead of its first client go-live.
Fixed plans and dynamic scheduling, side by side
| A fixed daily plan | Dynamic scheduling | |
|---|---|---|
| When the plan is made | Once, before the day starts. | Continuously, as the day unfolds. |
| An urgent job arrives | A planner phones round to find someone and reshuffles the rest by hand. | The plan is recalculated around it. |
| A visit overruns | Later jobs slip, and customers find out when the engineer is late. | Later jobs are re-planned before they slip. |
| Who is considered | Usually whoever is free and nearby. | Every available engineer, every time. |
| The planner's job | Re-planning all day. | Deciding the exceptions and handling customers. |
Why it was impractical before AI
Scheduling engineers is a large combinatorial problem. Each job has a location, a time window, the skills and qualifications it needs and sometimes parts; each engineer has a day already partly filled, a location and a set of qualifications. A good planner can balance a few dozen of these in their head. Past that, the plan becomes a compromise.
Rule-based scheduling tools helped with the first plan of the day, but rules break when the day changes in a way nobody wrote a rule for. AI can weigh every option again whenever something changes, which is what turns scheduling from a morning task into a continuous one.
What any dynamic scheduling tool has to weigh
Whatever the vendor, a dynamic schedule is only as good as what it takes into account. At minimum:
- Skills and qualifications. The engineer who arrives has to be able, and allowed, to do the work.
- Availability. Shifts, existing jobs and time off.
- Location and travel. Real travel time between jobs, not straight-line distance.
- Priority and deadlines. Emergencies and service-level deadlines first.
- Access and customer windows. When the site can actually be visited.
Where people stay in charge
A dynamic schedule still needs people. In a well-run one:
- Planners see the proposed changes and make the call on anything unusual.
- A visit the customer has already agreed is moved only for a good reason, and the customer hears first.
- Anything safety-critical goes to a person, every time.
Questions to ask a dynamic scheduling vendor
- Does it re-plan during the day, or only overnight?
- Does it check skills and qualifications before it looks at travel?
- Can it schedule subcontractors as well as employed engineers?
- Does it work with the CAFM or job system you already run?
- Can your planners override it, and can it explain its choices?
- How long does it take to go live?
For how AI fits the rest of the FM back office, see AI in facilities management, and for how it connects to your systems, connecting AI to your CAFM.