Case study · Public sector
Service requests & crew dispatch

From a text message to a finished work order.

A First Nation government in British Columbia ran its field operations on phone calls, paper forms, and memory. We extended the platform that runs its back office into a dispatch system — community requests arriving by web, email, or SMS, an AI agent filing them, and crews dispatched and closing work orders entirely over text message.

Sector
First Nation government, BC
Engagement
Platform extension · in production
Commercial
Fixed scope & price
Data & hosting
Canadian infra · client-owned
The challenge

The back office was systematized. The field wasn’t.

Community members reported problems by phone call, drop-in, or paper form. Work orders lived on paper and in memory. Once a report was made there was no reference number to follow up on, no record connecting the job to the crew that did it, and no way for the office to know what was still open — or what any of it cost.

Off-the-shelf field-service software assumes dispatchers at desks and crews running the vendor’s app on company phones. This department’s crews live in text messages, its requests come from a whole community, and some of its work is done for other departments and has to be billed back. It needed a system shaped around that — and it already had the platform to build it on.

The core change
Requests used to live in phone calls and memory. Now every report becomes a numbered, tracked record — and crews close the loop over ordinary text messages, with no app to install.
What we built

A dispatch loop the whole community can reach.

Four intake channels, one queue, and a crew workflow that runs on the phones people already have.

01

Four ways in, one queue

A community member can report a problem through a public web form — no login, no app — by email, at the office counter, or with a plain text message. Every channel lands in the same queue as a numbered service request.

02

An AI agent answers the texts

Text the department’s number and an AI agent files the request conversationally, hands back its reference number, and can say where it stands. Its powers are deliberately narrow — it can file and read, never assign, approve, or close — and every conversation is kept as a transcript.

03

Requests become work orders

The office reviews each request and promotes it to a numbered work order — scheduled, assigned to a crew, and tracked through every stage to completion. Categories, crews, rates, and even the AI’s instructions are dashboard settings the department controls itself.

04

Crews run on text messages

An assignment arrives as a text. The crew texts back when the job is done — hours, notes, even photos from the site — and the system completes the work order and automatically tells the person who reported the problem.

05

Emergencies interrupt, quiet jobs resurface

A request flagged as an emergency pages designated staff the moment it arrives. Jobs that sit idle trigger a nudge, and a Monday-morning digest lists everything still open.

06

Every job knows its cost

Work orders link to invoices and job costs on the same platform that runs the back office. Work done for other departments rolls up into a billed-back report — cost recovery as a printout, not an argument.

What changed

A request can’t get lost

Every report — texted, emailed, or handed over the counter — becomes a tracked record with a reference number, and the requester hears back automatically when the work is done.

The office sees everything

Open requests, work orders, who’s assigned, what’s stalled, and what got done — one live queue instead of paper forms and memory.

Crews without an app

The whole crew loop runs over ordinary SMS — nothing to install, nothing to learn. A crew member needs a phone that can text, which is every phone.

Billed-back work is recovered

Jobs done for other departments are tracked and reported for billing back — recovered revenue that used to be absorbed, tied to the true cost of each job.

Why it holds up to scrutiny
Governance
Roles, enforced approvals, and a complete audit trail on every action.
Data residency
Hosted in Canada. The data stays the client’s, under the client’s control.
Assurance
Automated tests and code review on every change before it ships.
Operated
We host, monitor, and maintain the system — model swaps included.
Have an operation that answers to a council, a board, or an auditor?
admin@xerad.ca
xerad.ca · Port Moody, BC
Xerad · AI & software engineering studio
Client identity withheld for confidentiality