AI in the City: Delivering Practical Benefits 

Exco Partners recently completed a Copilot study for the City of Melbourne who provides hundreds of essential services to businesses and people who live, work, and study or socialise in the municipality, and millions of other stakeholders.

The challenge and opportunity

The City of Melbourne has several internal support functions (HR, IT, Procurement, Finance, Legal, Risk & Governance) that support service delivery. These functions are facing common challenges:

• Employees don’t always know where to go to get assistance, what process to follow
• With multiple systems and distributed knowledge bases, people have trouble accessing required information
• Employees seeking support and information outside of the 9-5/Monday to Friday schedules.
• Increasing workloads
  

Generative AI provides the opportunity to reduce the amount of time allocated to lower-value tasks, enabling increased focus on value-added activities aimed at improving customer and/or employee experiences.

To explore and understand the business improvement potential of Generative AI technology, the council initiated and scoped a two-part pilot project utilising Microsoft 365 Co-pilot and Co-pilot Studio.

The objectives of the pilot were to understand:

• Risk profile (readiness assessment)
• User uptake and adoption across representative archetypes
• True cost of ownership (based on users, transaction types and volumes)
• Business benefit (return on investment & employee experience)

Exco Partners was engaged to develop, test, and launch pilot solutions, support the pilot user base, and measure pilot outcomes. The pilot solutions needed to be
designed such that, should pilot outcomes be positive, they can be scaled to production.

The approach

The plan for the pilot included three main phases with both streams running in parallel.

Phase 1 – Discovery: This phase enabled Exco Partners and the client to refine the pilot objectives and scope, success measures, measurement framework and target use cases. This period was also used to assess the organisation's readiness using a risk-based framework.

Phase 2 - Build, prepare, train: This phase was used to build and configure the Co-pilot Studio Solution and to prepare the pilot participants through training and user adoption activities. The team prioritised, fit-for-purpose, effective training that was made available to all pilot participants to ensure:

• Strong familiarity with the piloted solutions
• Intuitive and easy-to-navigate support features
• Development of effective prompting techniques to optimise outcomes

Training covered how to use Microsoft 365 pilot and Copilot studio, as well as principles of Responsible AI use, limitations in the organisation’s environment and associated risks.

The Solution

Phase 3 - Run: The pilot included two streams running con-currently:

• Stream 1 - provided access for employees to test and learn the capabilities Microsoft 365 Co-pilot.
• Stream 2 - provided access for employees to the Co-pilot Studio solution to understand the capability of a Gen AI-based agent.

This phase included four two-week pilot sprints. Each sprint included a kick-off to communicate the sprint objectives and report the outcome of the previous sprint, group check-ins and one-on-ones and constant communication and support.