Case studies

What these businesses stopped handling manually.

Three workflows in production. One is measured from the client’s own records. For the other two the client has not released figures, so you get the workflow instead.

Show us where it breaks

Real deployments. No demos presented as client results.

In production

Three production workflows. Three different levels of evidence.

Exterior cleaning

Two locations. First went live 2025.

A customer journey built onto ServiceM8, not instead of it.

Inbound calls answered, callers matched against the records that already exist, quoted from the client’s own prices, and booked. All of it inside ServiceM8, which stayed the system of record throughout.

258 enquiries created from AI-handled calls. 125 progressed to priced work, $135,716. 68 completed and invoiced.

Both locations of an exterior-cleaning client combined, pulled from the client’s own job records on 9 August 2026. Enquiries originated through AI-handled inbound calls.

Read the full case study →

Multi-service trade

Custom pricing backend.

When the software couldn’t do it, we built around the software.

Was Quoting needed a person to interpret the request, look up the property and the pricing, then work out the charge before anything could go out.

Now The request is read, priced against the client’s own pricing engine, and drafted ready to go. Anything the rules do not cover stops and waits for a person. The engine is theirs, so the answer matches what they would have quoted by hand.

The client has not released commercial results. What we can show is the production workflow: the request is interpreted, the client’s own pricing engine returns the figure, and anything it cannot price stops for review.

See the quoting workflow →

Insurance advice

Adviser pilot. Third-party quoting API.

Three trips through the same form, down to one.

Was Comparing cover meant entering the client into the quoting platform, re-typing the same details for every variation, changing one figure and doing the whole thing again, then going back to the provider’s own portal to produce the official document.

Now The scenarios are entered once and quoted straight against the platform’s API. Change one detail and only the affected scenario re-runs. Every quote keeps its history, so an earlier version is still there after the client changes their mind.

Quote set · 3 scenarios Rev 4
Rev 1–3 kept and reopenable. Illustration, not a screenshot.

The adviser chooses the provider and submits the application. The system compares and never recommends. Pilot stage, no outcome figures.

We never combine figures across workflows.

Also in production

Quieter deployments, same discipline.

  • Locksmith and emergency trades Answering since May 2025, at hours a person was never going to cover.
  • Self storage Unit holds are gated at the tool level. The agent cannot place one outside the rules, and every gate has a test behind it.
  • E-commerce operations Order, supplier, logistics and customer communication run as one sequence instead of four separate jobs.

Which of these is closest to yours?

Show us the workflow you want off your plate.

  • Maybe none of those three look anything like your business.
  • Maybe the shape is the same and every detail is different.

Those workflows are worth more as shapes than as stories. Show us yours and we’ll tell you which one it resembles, what we’d expect to be different, and what we wouldn’t promise until we had seen your rules.

Would rather talk it through than type it out? Book a 30-minute call.

About the process

Four questions. Owen picks up the rest on the call.

Add a bit more (optional)

Owen reads every submission. You’ll usually hear back within one working day.