AI adoption in professional services: beyond the dashboard

The approach

There was clear alignment from the outset that the human side of change was key. Karlee joined as AI Advisor, a trusted partner whose thinking complemented that direction and helped translate it into a practical engagement rhythm. That shared view meant work could move quickly, with limited friction and no need to re-establish the principles underpinning the change.

Sessions were structured to build momentum over time rather than delivered as isolated events.  Each focused engagement was followed by regular drop-ins where participants could share how they had been using Copilot that week and ask questions in a low-pressure environment.  Playback sessions brought teams together to surface what was being discovered in practice, creating visibility across departments. Themes from each session shaped what came next. The learning loop was built in by design.

A Copilot Community was established on Viva Engage to extend the conversation between sessions. Rather than routing individuals toward private support, the community became a space where questions and insights were shared openly, creating patters that informed direction of subsequent work and reducing the isolation that often characterises early AI adoption. 

What the numbers don’t show

One of the clearest signals of success was what was happening towards the end of the engagement. Conversations that had started within one team began creating natural connections across departments, with colleagues recognising that what was solving a problem in one area was directly relevant to their own.  People were applying Copilot in ways that went beyond the basics, and doing so without being prompted.

Usage metrics alone don't tell that story.  High daily active use can still mean people are reaching for a tool the way that they would a search engine: familiar enough to open it, not yet taking advantage of it.  What matters is whether the behaviour underneath is shifting, whether people are collaborating differently, solving more complex problems, and beginning to reach for AI as a first instinct.  At Mallesons, usage remained strong, but the more significant signal was the nature of how people were working with Copilot had begun to change.

The foundations for intelligent agent development

As the engagement progressed, Karlee balanced advisory input on AI agent design with hands-on experimentation inside the Microsoft ecosystem. Working within the firm's existing toolset and governance constraints, she helped the team test what was possible in practice while providing recommendations on agent architecture to improve output reliability and reduce the risk of hallucination or inconsistency. In contexts where regulatory obligations are significant, including client onboarding processes subject to anti-money laundering requirements, those early design decisions carry long-term consequence.

The engagement also surfaced important governance considerations for future agent development. The opportunities identified were documented in a structured way, capturing their purpose, risk profile, intended user group, data requirements, and process context, enabling a clearer base to take forward and keeping accountability visible as the program scales.

The outcome

Mallesons made the decision to procure Copilot licenses across the operations team.  That procurement was a milestone, not a measure.  Organisations investing in AI will see that investment reflected not in the license numbers, but in whether the people using those tools are actually working differently.  At Mallesons, that shift was visible: people using AI tools thoughtfully, sharing what they were learning, building on each other's practices, and sustaining that momentum.