Share Case Study
Case Study
UniSuper is an industry super fund with strong long-term performance. With workplace technology evolving at a rate of knots, it recognised that it was a strategic priority to become an AI fluent organisation with leading tools in place and embedded responsibly in everyday work.
To meet this challenge, the People team partnered with Nous to develop a practical AI Agent Playbook designed to help employees design, build and use AI agents in their everyday internal work. The playbook gives UniSuper people clear guidance on how to integrate AI capabilities into their roles effectively and responsibly. Key requirements for the project included:
- Delivery within tight timelines. The playbook needed to be fully developed within a four-week delivery window.
- Customisation for specific tools. The content had to focus on Microsoft 365 AI agent tools, such as Agent Builder and Copilot Studio, ensuring it met the precise technological requirements of UniSuper.
- Inclusive co-design process. The development process had to actively involve employees with varying levels of AI expertise, ranging from novice users to those with advanced technical skills.
- Governance, risk & compliance. The playbook needed to align seamlessly with UniSuper’s AI policy and governance framework, providing employees with clear guidance to identify potential risks and supporting controls and when and how to seek support.
“Helping UniSuper get ahead of a technology that is still finding its feet, but is clearly going to become central to how organisations work, was a really exciting opportunity,” says Director Ed Verbeek
“This wasn’t just about producing a document. It was about building real capability,” says Miranda Hewitt, Head of People Solutions at UniSuper. “We’ve launched the playbook and are already seeing people use AI in practical ways that save time and improve how work gets done.”
Working at speed and scale
The project posed several challenges. The four-week delivery timeline required an agile approach focused on the elements most critical to success. The materials also needed to be accessible for employees new to AI while still useful for more advanced users. At the same time, the playbook needed to reflect UniSuper’s internal AI policies without becoming overly procedural.
“The challenge as we saw it,” says Verbeek, “was driving genuine AI agent adoption across UniSuper's entire workforce.”
“Striking the right balance between practical guidance and approachability, without oversimplifying the content, meant the design of the playbook was just as important as what it said.”
To respond to these challenges, the team used a structured and collaborative implementation approach. This enabled the playbook to be developed quickly while meeting the needs of UniSuper’s diverse workforce.
Our three-phase iterative process
The project was designed around three key phases to achieve the specified outcomes on time and to an exceptional standard:
1. Establishing an anchoring framework
Before engaging directly with employees, we created a robust foundation for the playbook in the form of a five-stage AI Agent Lifecycle Framework. This framework was developed to cover the entire journey of an AI agent, from ideation to deployment and ongoing monitoring and improvement.
The framework was presented and tested with UniSuper’s working group during the first of three co-design workshops. Employees were encouraged to share their feedback and provide valuable input based on their current experiences, needs, and aspirations regarding AI at work. This exercise validated the framework as a strong foundational structure to organise the playbook content.
2. Prioritising guidance materials
In Workshop 1, we collaborated with employees to identify key pain points and most valuable areas for guidance within each stage of the lifecycle. This process ensured that the playbook would focus on content that provided the most direct impact on employees’ daily workflows.
By prioritising the most practical and immediately actionable sections, we not only ensured relevance but also created a resource that employees could reference with confidence when engaging with Microsoft 365 AI tools, such as Agent Builder and Copilot Studio.
3. Iterative build-and-test sprints
Following the groundwork laid in the first workshop, we adopted an iterative methodology to rapidly develop, evaluate, and refine content during the subsequent workshops. Each of the lifecycle’s five stages was reviewed and expanded systematically.
- Material development. We drafted detailed guidance and resources tailored to the specific stages of the lifecycle identified in the framework.
- Hands-on testing. During the workshops, participants used the materials in structured exercises to design and test AI agents within the Microsoft 365 platform. This process revealed valuable insights into practical use cases and the challenges employees might face.
- Feedback gathering and refinement. Workshop participants shared live feedback as they engaged with the draft materials. They highlighted areas where the guidance was clear and actionable, as well as instances where additional clarity or content updates were needed. This feedback was incorporated into the materials between workshops, ensuring continuous improvement and alignment with business needs.
The iterative process allowed the AI Agent Playbook to evolve rapidly into a practical and accessible document tailored to the real-world needs of UniSuper employees.
Building capability into the future
The AI Agent Playbook provides UniSuper employees with a comprehensive, user-friendly guide to responsibly and effectively design, deploy, test and manage AI agents as part of their routine work tasks. It provides step-by-step assistance, designed to help employees:
- Identify relevant use cases for AI agents within their roles.
- Identify triggers and escalation points for higher risk use cases
- Build AI agents using Microsoft 365 tools such as Agent Builder and Copilot Studio.
- Integrate AI agents seamlessly into their workflows where appropriate, with human review and relevant controls in place.
- Monitor the performance of AI agents to ensure they are operating as intended.
- Align their use of AI tools with UniSuper’s broader AI policies and governance structures.
The playbook’s emphasis on governance ensures employees understand when and how to seek support and approval when using AI agents. This supports the organisation’s commitment to maintaining ethical, transparent, and effective use of AI technologies.
This represents an important step in UniSuper’s broader AI journey. It gives the organisation a practical foundation to continue building workforce capability and realise the benefits of AI in day-to-day work. It also marks a significant step forward in UniSuper’s journey towards becoming an AI-first organisation.
“We’ve seen a clear shift in both confidence and capability. People are moving from curiosity about AI to using it in their day-to-day work,” says UniSuper's Miranda Hewitt.
What you can learn from our work with UniSuper
Involving employees directly in the development process helps create invested advocates who understand the tool, believe in its value and can support adoption across the organisation.
Anchoring the playbook in a clear lifecycle framework before developing the guidance gave the project direction and helped avoid scope creep under tight timelines.
For a document designed to shift behaviour, style and communication matter as much as content. If the tone is too technical, too dry or too generic, the guidance is less likely to connect.
Key people in this project
Tony Fiddes
Principal
Tony has expertise building organisational performance through people strategy, leadership development and culture change.
