Field service businesses have a structural characteristic that makes them unusually good candidates for AI operations work: the expensive resource is a technician's time in front of a customer, and an enormous amount of it is consumed by things that are not that.

Four patterns recur across every trade we have worked with.

1. Intake and dispatch

The call comes in. Someone captures it, determines urgency, checks who is available and appropriately certified, considers where they are, and schedules. Done well it takes several minutes and considerable local knowledge. Done at volume, during a storm week, it becomes the constraint on the entire business.

What changes: capture, classification, enrichment and routing happen automatically; the dispatcher handles exceptions and judgment calls. The measurable outcome is time to a scheduled appointment, which for most operations drops from hours to minutes.

The trap: automating the acknowledgment without the routing. An instant reply followed by silence is worse than a slower human response.

2. Scoping and quoting

The estimate is the bottleneck on revenue in most field businesses. Quotes go out slowly and inconsistently, and the ones that go out slowly lose to the competitor who quoted same-day.

What changes: a structured scope is generated from the request and site information within minutes, reviewed and adjusted by a person, and sent same-day. Consistency across estimators improves margin as much as speed improves close rate.

The trap: scoping without confidence indication. A quote that presents a solid number and a guess identically is how jobs lose money.

3. Parts and truck stock

The second visit is the most expensive event in field service. It happens because the part was not on the truck, which happens because nobody predicted what this job would need.

What changes: the system predicts likely parts from the described symptom, the equipment history at that site and comparable past jobs, then checks stock and flags the gap before the technician leaves.

The measurable outcome: first-visit resolution rate — arguably the single most important operational metric in the trade, and one many companies do not track.

4. Follow-up and recurring revenue

Maintenance agreements, warranty follow-up, seasonal service, the callback on the job from eighteen months ago. Everyone knows this revenue exists. It goes uncollected because nobody has an afternoon.

What changes: the follow-up happens on schedule, in the company's voice, with the specific context of what was done and what is due. A person approves the outbound.

The trap: generic blasts. "It's time for your annual service" is ignored. "Your system was last serviced in April and the compressor showed early wear — worth checking before summer load" gets a response, and it requires the system to actually know the job history.

What we do not recommend automating

The conversation at the door. Explaining a repair to a worried homeowner. Pricing negotiation. The judgment call about whether to repair or replace.

These are the parts customers remember and the parts that determine whether they call you again. The point of automating the other four is to give the technician more time and better information for these.

Where to start

Whichever of the four is currently your constraint — and that is usually obvious to whoever runs operations. Do not start with the most interesting one. Start with the one that is capping revenue, measure it before and after, and expand from evidence.