M3 Property

THE CHALLENGE

M3 Property had rolled out Microsoft Copilot Premium licences across the firm, with CEO Andrew Duguid leading the push. Leadership wanted to move the firm from AI literacy to measurable business outcomes, with valuers and support staff spending more of their week with clients and less of it on admin.

Banks and other clients rely on M3's valuations, so any use of AI had to keep accuracy and compliance intact and keep the valuer accountable for the final opinion.

M3 asked Collective Campus to run a hands-on AI Design Sprint inside its annual staff conference on the Gold Coast. The goal was for staff to find the problems in their own workflows, redesign those workflows with AI, build working prototypes, and show leadership what they built.

WHAT WE DID ON THE DAY

Steve Glaveski led the day with co-facilitator Tessa Kowaliw. M3 staff worked at 20 tables of six.

• Foundations. We opened with design thinking, lean startup, customer interviewing and jobs to be done, so teams had a shared method before they touched any tools.
• Problem discovery. Each table picked one process or problem to focus on, mapped how the work happens today on a wall canvas, and marked the steps that cost the most time. M3 had gathered pain points from across the firm before the day, covering valuations and reporting, business development, quotes and admin, finance and work in progress, risk and compliance, and HR.
• Ideation. Teams turned their problem into a how might we question, generated ideas, and chose one to build.
• Prototype sprint. Each table split into two groups of three. One group built a working prototype in Copilot while the other built the pitch. Our rule for the session was to redesign the workflow first and then apply AI, so teams avoided layering a tool on top of a broken process.
• Guardrails. Valuers stay accountable for the opinion, sign-off stays human, and client data stays inside approved tools.
• Shark Tank. Teams pitched in about three minutes each to a panel of M3 leaders, joined by Steve and Tessa. The panel scored each pitch on the problem, the solution, the impact in time or cost, and the demo.

RESULTS

Teams built working prototypes in Copilot on the day. Across the room they produced 20 ideas. Eight teams pitched to the judges, and the panel picked two winners.

Many of the prototypes will go to M3's AI steering committee for further prioritisation, development and implementation.

The ideas clustered around five problems that came up again and again across tables: getting new jobs ready to start, capturing sales and leasing evidence once so it can be reused, moving from inspection to a draft report faster, seeing work in progress across offices, and making M3's own knowledge and client history easier to find.

One team proposed cutting repetitive transcription and model setup in valuation work, then testing it by comparing a completed model against manual preparation and measuring total time, input accuracy and corrections.

The steering committee gives M3 a clear route from AI literacy to measurable business outcomes. Staff designed the prototypes around real workflows and built them with tools the firm already pays for, and the committee can now prioritise, develop and measure them.