Skip to content
Field services · 14 weeks

A dispatch system that survived the August heat wave

Constraint-aware scheduling and an offline-first technician app replacing a whiteboard, three spreadsheets, and a group text — built and live in fourteen weeks.

Micro-SaaS applicationsAI automationCloud & platform
A field technician holding a rugged tablet showing a job checklist
+22%
Jobs completed per tech per week
5% → 1.4%
Return visits for wrong parts
2 hr
Customer arrival window
100%
Technician adoption by week two
The challenge

Where they were

Engagement
Duration14 weeks
Team3 engineers
IndustryField services

A commercial HVAC contractor with 140 technicians scheduled every job from a dispatcher's memory.

One in five visits ended without the right part on the truck, and emergency calls destroyed the day with no way to re-plan.

Three field-service platforms had been evaluated; none could express the certification and equipment constraints the work actually had.

A handover training session with a presenter and colleagues taking notes
Approach

What we did

01Encoded the constraints the dispatchers carried

Certifications, truck inventory, drive time, customer SLA, and the informal rules about which techs work well at which sites — captured explicitly, then used by the scheduler.

02Predicted parts at booking time

A model trained on eight years of job history surfaces likely parts by equipment model when the call is taken, so trucks are stocked before they leave the yard.

03Designed the mobile app around gloves and bad signal

Offline-first, four taps to close a job, and voice notes that get structured into the work order. Adoption was 100% in week two because it was faster than the paper it replaced.

04Re-optimized when reality intervened

Emergency calls trigger a re-plan of the remaining day with the smallest set of changes, and affected customers get an updated window automatically.

During the August heat wave we ran 40% above normal call volume with the same crews. The old whiteboard would have simply fallen over.
Director of Field OperationsCommercial mechanical contractor
Stack

What it was built with

TypeScriptReact NativePostgresPythonTwilioQuickBooksGCP
Similar situation?

Let’s talk about yours.

Mention this case study in your note and we’ll come to the call with the specifics of how it went — including what we’d do differently.

We use your details to reply to this request only. No sequences, no list.

Taking two new engagements this quarter.