Your Dispatcher Is Solving a Math Problem by Hand Every Morning

It’s 6:45 AM. Your dispatcher has 41 jobs, 11 techs and a coffee going cold. Two techs can do gas work. One is the only person you trust on commercial refrigeration. Three trucks have the 40/5 capacitors, and one doesn’t have any. Mrs. Alvarez was promised 8 to 10 and has called twice about it. A member with no cooling has to go first.

By 7:30, the board is built. By 10:15, it’s wrong. A drain job that should have taken an hour is in its third hour, a water heater turned into a repipe, and a tech called in with a sick kid. The dispatcher starts again, on the phone, from memory, while the CSRs ask where every truck is.

I should be clear about what I bring. I’ve never dispatched a truck. For almost 18 years I ran a digital agency, Infographic World, and today I use AI to fix real problems inside businesses. I know how work moves through an office, and dispatch is where a whole shop’s day either flows or jams.

Where the time actually goes

Building a board means weighing all of this at once, for every job:

• Skills and licenses. Who’s qualified for gas, for panel work, for commercial equipment, for the new heat pump line.
• Location. Where each tech starts, where each job is, and what the roads look like at 8 AM versus 2 PM.
• Parts on the truck. A tech who has to stop at the supply house can burn a good part of the morning before the job even starts.
• Promised windows. What the customer was told, and how hard it is to move.
• Priority. Members first, no-cooling and no-heat before tune-ups, the job the owner promised a friend.
• Job duration. The biggest unknown.

Then the day starts, and the plan meets the real world. A job runs long, the next three windows slide, and the dispatcher has to pick who to call, who to move and what to tell the customer. The cost shows up in three places. Windshield time, as techs cross town to make the board work. Missed windows, which turn into angry calls and canceled jobs. And overtime, when the last job of the day runs late.

What AI-assisted dispatch looks like

It proposes the board

The night before, or early in the morning, the AI drafts a board. It reads the jobs, the techs, their skills, their start locations, what’s on each truck and the promised windows, then lays out a plan that keeps drive time down without breaking any of your rules. The dispatcher sees the proposal, not a finished decision, and moves whatever needs moving.

It estimates duration from your own history

This is the quiet win. The AI looks at your own completed jobs: job type, equipment, age of the home, which tech, and how long it actually took from arrival to done. So a “drain clear” in a 1960s house with cast iron gets a different estimate than one in new construction. Better estimates mean fewer windows broken by noon.

It watches the day and flags windows at risk

As techs update their status, the AI keeps checking the board. When a job runs long and the 1 to 3 window down the line is now in trouble, it tells the dispatcher before the customer finds out. It suggests a fix, like swapping two jobs between techs who are near each other, and the dispatcher decides.

It texts the customer before they call

Once the dispatcher approves, the customer gets a short, honest text: the tech is running about 40 minutes behind, here’s the new estimate, reply if that doesn’t work. That one text heads off a lot of “where’s my tech” calls, and it lets the customer reschedule instead of waiting at home.

The dispatcher stays in charge throughout. They know which customer only wants Mike, which tech is having a rough week, and which “simple” call is really a replacement waiting to happen. The AI handles the arithmetic so the dispatcher can make those calls.

Run the numbers

There are two parts to the formula. First, the drive time: windshield minutes saved per tech per day, times the number of techs, times your loaded labor cost per hour. Second, the capacity: one extra job per tech per week, times your average ticket.

For a reference point on pay, the BLS lists 2025 median pay for HVAC mechanics and installers at $29.33 per hour. Your loaded cost is higher once payroll taxes, insurance, benefits and the truck are in, and your own number is the one to use.

Say, for the sake of the math, a shop runs 10 techs at a loaded cost of $55 an hour, and a better board saves each tech 25 minutes of driving a day. Those inputs are made up.

25 minutes × 10 techs = 250 minutes, a bit over 4 hours a day. At $55 an hour that’s about $230 a day, or roughly $57,000 over 250 working days.

Now the bigger lens. Twenty-five minutes a day is about two hours a week per tech, enough for one more service call. Say each tech fits one extra job a week at a made-up average ticket of $350, across 50 weeks.

10 techs × 1 job × $350 × 50 weeks = $175,000 a year in added revenue.

To run it on your own business, pull three numbers:

1. Drive minutes per tech per day. Your GPS or the time between “on the way” and “arrived” in your field service software, averaged over a month.
2. Loaded labor cost per tech hour. Wage plus taxes, insurance, benefits and truck cost, from your accountant.
3. Average ticket on a service call. From your own closed jobs.

Then decide honestly how many saved minutes could become a booked job, given your demand and season.

Picture a Monday, before and after

This is a made-up scene. Picture a 10-truck plumbing and HVAC shop on a Monday in July.

Before. The dispatcher builds the board from scratch at 6:30. By 9:45, two jobs are running long and four windows are at risk. The dispatcher finds out when the customers call. Two techs cross town to cover, one with a stop at the supply house for a part. The CSRs take 14 “where’s my tech” calls before noon. Three techs log overtime.

After. The dispatcher opens a proposed board at 6:30, moves three jobs for reasons the system couldn’t know, and approves it by 6:50. At 9:20, the system flags that a drain job is running long and the 11 to 1 window is at risk. It suggests swapping two jobs between techs who are eight minutes apart. The dispatcher approves, and the affected customer gets a text before they’ve thought to call. The CSRs take four “where’s my tech” calls. One tech logs overtime.

What has to be in place first

• Job history with real timestamps. Arrived and completed times on past jobs, with job types used consistently. This feeds every duration estimate.
• A written skills list. Every tech, what they’re qualified and licensed to do, and what they prefer not to.
• Truck stock you can see. Even a rough list of what each truck carries beats nothing.
• Your window rules. How windows are promised, how much slack they get, and who goes first.
• Techs updating status from the field. On the way, arrived, done. Without it, nothing can watch the day.

Guardrails worth keeping

• The dispatcher approves the board and every change. The system proposes. People decide.
• Customer texts use approved wording. Timing updates only, written in advance, sent after the dispatcher okays the change.
• Be open with techs about location data. Tell them what’s tracked and what it’s used for. Use it to plan the day, not to build a case against someone.
• No pressure to rush. Duration estimates are for planning. A tech who needs more time on a gas line or a panel takes it.
• Emergencies override everything. A gas smell or a flooded basement goes to the top of the board, whatever the math says.

How you’ll know it’s working

• Drive minutes per job, averaged weekly.
• Jobs completed per tech per day.
• On-time arrival rate within the promised window.
• “Where’s my tech” calls per day.
• Overtime hours per week.

Where I’d start

Start with duration estimates. Pull a year of completed jobs, have the AI estimate how long each job type takes in your shop, and show those estimates next to the dispatcher’s own guesses on the current board for two weeks. No changes to how the board gets built.

If the estimates hold up, they make every board better, by hand or by proposal. If they don’t, your job data needs cleaning first.

If you want to see what that time is worth in your shop, the AI Opportunity Calculator is a quick way to run it.