Case study
AI defect quoting for a national fire protection provider
A national fire protection services provider with branches across Australia asked whether AI could speed up turning technicians' defect records into customer quotes. We ran discovery and design, then built an engine that checks and categorises each defect and compares its result with what the quoting team does.
The short version
Who
A national fire protection services provider with branches across Australia, a field technician app, an ERP and a separate quoting system.
The problem
Defects found on site arrive in the office with uneven detail, so quoters spend time chasing technicians and rekeying before a quote can go out. Every day a defect sits unquoted is work not yet booked and a customer waiting.
What we did
Discovery and design on production data, then an AI engine built to run inside the client's own cloud in Sydney.
Where it is now
The engine is built and tested, and the next stage runs it in shadow mode beside the quoting team before anyone relies on it.
How the work ran
Discovery on real data
Read-only access to production records showed which defect records were complete enough to quote and which went back to the technician.
Design with their IT team
The architecture was agreed with the client's IT and data leads: dedicated read-only views, a restricted login and the AI model in their own cloud.
Build the engine
A web service that picks up new defects, categorises each one and suggests the quote lines, with personal details stripped out first.
Shadow first
The team quotes as normal and sees no AI output. Once a day the engine compares its result with what the team did, and agreement is tracked per category.
Nothing leaves their boundary
Everything runs on the client's infrastructure in Australia. The engine reads through restricted views, and the AI provider does not train on the data. Each result is logged, so the project team can see what the engine decided and why.
- Hosted in the client's own cloud in Sydney
- Read-only access through restricted views
- Each result logged for review
A person stays in charge
Quoters keep the final say. Shadow mode measures accuracy on real work before anyone sees a suggestion, and the business owner decides when to move to co-pilot, using the agreement rate as the evidence. The pilot covers chosen territories and customers, and the engine ignores everything else.
- Shadow mode before co-pilot
- Go-live based on measured agreement
- Pilot limited to chosen territories and customers
Questions
Can AI turn technician defect reports into quotes?
Yes, when the defect records carry enough detail. AI can categorise each defect, flag records that need more information from the technician, and suggest the quote lines from past quotes, with a quoter approving every quote before it goes to the customer.
What is shadow mode in an AI rollout?
The AI runs on live work in the background while staff carry on as usual. You compare its answers with the decisions your people made and get an accuracy figure for each category before anyone relies on it.
How do you keep field service data private when using AI?
Host the AI in your own cloud in Australia, give it read-only access to the data it needs, remove names and contact details before each AI call, and choose a provider that does not train on your data. Log every result so your team can audit it.
How long does an AI quoting pilot take for a fire protection or field service business?
Discovery and design usually takes a few weeks on real data. The build follows, then several weeks of shadow running on live work before co-pilot. Access to systems and time from your IT team usually set the pace.
Related
Field service and facilitiesJobs, assets and compliance evidence.
Custom AI developmentAI that reads, checks and drafts the first pass.
Private AI infrastructureAI that runs inside your own environment.
Our work, stated plainlyClient delivery, and products we built and run ourselves. Each one says exactly what it is.
Quotes stuck between the field and the office?
Book a free 30-minute review and we will show you where AI fits.
Fixed-fee phases, agreed before we start. Stop after any phase.