Operations · Process breakdown
Let rules build most of the weekly plan and confirm slots with customers, while planners keep the exceptions and same-day changes.
Field service scheduling is usually one planner, a whiteboard in their head, and a phone that never stops. Most of field service scheduling is nonetheless rule work: jobs have durations, technicians have skills and availability, sites have addresses, and contracts set priorities. Rules can build a plan from that and confirm it with customers. AI helps only where the input is messy, such as a customer reply with a gate code and a “not before ten”. And people stay in charge of the breakdown at 8:15 that upends the day. The first thing to fix is job data: without durations and skills, no schedule is worth trusting.
Every job enters the queue with an estimated duration, the required skills and parts, the site address and the contractual deadline. Jobs missing any of these are held back, not planned on a guess.
A deterministic planner assigns jobs by skill, availability, travel time and priority, within agreed limits such as maximum travel per day and buffers between jobs. The same inputs always produce the same plan.
Customers receive proposed windows by email or text and confirm or pick an alternative. Confirmations update the plan; no reply after a set time triggers a call task.
Replies and notes such as access instructions, parking, on-site contacts or safety requirements are turned into structured fields the planner can use. AI does the reading; the constraint itself is stored as data.
The planner checks the flagged items: overtime, skill compromises, split visits, VIP customers. AI can summarise why each flag was raised, the planner decides.
Breakdowns, sick technicians and angry customers need judgement and a conversation. The system can show options and consequences; the dispatcher chooses and owns the promise made to the customer.
Plan one region or team with the rule-based schedule for two weeks while the planner keeps working as usual. Count the overrides the planner makes and why, travel time per job, missed or moved slots, and the confirmation rate. Use the overrides to fix rules and data before widening the rollout.
Buying field service scheduling software before durations and skills are recorded. Optimisation on bad data produces confident plans that fall apart by Tuesday.
Route and assignment planning is an optimisation problem with clear rules, so deterministic planners handle it better and more predictably than a language model. AI is useful around the edges: reading free-text constraints, summarising why a plan is tight, or drafting the message to a customer whose slot moved. Keep the planning itself rule-based and explainable.
Everything with a trade-off the rules do not capture: whether to accept overtime for a key customer, whether a less experienced technician can take a complex job, how to handle a breakdown that collides with contractual deadlines. Automation should surface these as flagged choices with consequences, and a person should make the call and speak to the customer.
Compare it with the manual plan on the same weeks: number of planner overrides, travel time per job, jobs completed on the first visit, slots moved after confirmation, and how often technicians ran out of time. If the automated plan needs constant correction, the data or the rules are wrong, and that is what to fix next.
Illustrative workflow guidance by Arcgent. Each business needs its own assessment. No integration or savings claim has been verified for your systems.
Get an assessment based on your own steps, systems, and constraints.
Audit my process See an example report ↗