Delivering AI with confidence: How PRINCE2 Agile turns experimentation into business value

Mike Boutel, Head of Training, PM-Partners and PRINCE2 Ambassador

Artificial intelligence has quickly become a strategic priority for organizations worldwide. However, many are still trying to figure out how to turn their investments into measurable business value. While AI tools are increasingly accessible, successful implementation requires more than just selecting the right technology. Organizations need to identify where AI can improve business outcomes, equip people with the skills to use it effectively and establish governance that supports experimentation while ensuring accountability. Without this foundation, AI initiatives risk becoming isolated pilots that fail to deliver lasting benefits.

Here, Mike Boutel, PRINCE2 Ambassador and Head of Training at PM-Partners, explores the challenges organizations face as they adopt AI, drawing on conversations with business leaders in Australia. He explains why AI initiatives should be approached as business change rather than technology projects, and how the principles of PRINCE2 Agile offer a practical framework for governing experimentation, demonstrating value, and scaling successful outcomes.

AI has quickly become part of everyday working life. Across organizations, employees are using tools such as Microsoft Copilot, ChatGPT, Claude and Gemini to draft documents, analyze information and automate routine tasks. Yet one message emerged consistently from conversations with business leaders at this year's AgileAus conference: organizations have invested heavily in AI, but many still struggle to demonstrate measurable business value.

A recurring question in discussions with organizations is straightforward: if you have 1,000 employees using multiple AI tools at a cost of $20 per user per month, what return are you seeing on that investment? Very few organizations can answer that question confidently. In many cases, AI has been introduced as a technology deployment rather than a business initiative, making it much harder to define, measure and realize value.

Start with the business problem

One of the biggest misconceptions about AI is that it is an IT project. IT has an important role in providing secure access to approved tools, but identifying where AI can improve the business is the responsibility of the people doing the work. They understand the processes, bottlenecks and opportunities better than anyone else.

Too often, organizations will deploy Copilot or another AI assistant, ask employees to complete a short Microsoft course, watch a few YouTube videos, or complete some LinkedIn Learning modules, and then assume productivity will naturally follow. In practice, simply giving people access to AI tools rarely leads to significant improvements in productivity.

Without the right education, people often don't realize what these tools are capable of. This makes it much harder to identify where AI can improve team workflows or remove repetitive tasks. The conversation should begin with the business, the workflow and the outcome you are trying to achieve. Once teams understand what AI can do, they are much better placed to identify opportunities, prioritize initiatives and begin experimenting.

Governance needs to keep pace

Governance remains another challenge. Many organizations have developed AI governance frameworks and implemented them throughout their operations, but the rapid evolution of AI often outpaces these frameworks.

To manage this uncertainty, some organizations have chosen to restrict AI use altogether until they feel more comfortable about the associated risks. Others approve only a limited set of tools. However, employees who want access to more capable models frequently find ways to bypass these restrictions, often using personal devices or consumer accounts outside of the approved governance processes.

Governance alone is not enough. Organizations also need to help people understand how to use AI responsibly within those guardrails while policies and education need to develop alongside the technology.

AI projects are still projects

Another common misunderstanding is that AI requires a completely different approach to delivery compared to other projects. In reality, the fundamentals remain the same. You still need a business case, funding, executive sponsorship, accountability for delivering benefits, and governance.

However, AI initiatives do involve more experimentation than many traditional projects, with pilots having mixed success rates. The challenge is avoiding what we call the “proof-of-concept graveyard” – promising pilots that never progress into projects delivering meaningful business value. Organizations need to identify which pilots demonstrate genuine value, stop those that do not, and scale those with the greatest potential.

Organizations also need to continually reassess whether building a bespoke AI solution remains the right approach. AI is evolving so quickly that a capability developed today may soon be available as an off-the-shelf application. Regularly asking whether it still makes sense to build rather than buy is part of the governance process.

How PRINCE2 Agile supports AI delivery

The principles of PRINCE2 Agile provide a practical framework for addressing these challenges. While AI initiatives need room to experiment, they also need the governance to ensure those experiments remain focused on delivering business value.

Each of the PRINCE2 Agile principles has a practical application for organizations introducing AI:

- Start with value: PRINCE2 Agile emphasizes the importance of continued business justification. Before selecting an AI tool, define the business outcome you want to improve. The technology should support your objectives, not become the objective itself.

- Learn as you go: AI pilots are learning opportunities rather than one-off experiments. The learn from experience principle encourages teams to capture insights on the use of AI, adapt use cases quickly, and apply their learnings to future initiatives.

- Be clear who owns decisions: While AI can automate tasks, generate recommendations, and accelerate decision-making, human judgement remains essential. PRINCE2 Agile’s emphasis on clearly defined roles and responsibilities ensures accountability ultimately rests with people.

Be clear about who owns decisions

- Take it step by step: The manage by stages principle encourages organizations to test, validate, and then scale successful initiatives, so organizations can demonstrate the value of AI incrementally before expanding successful approaches.

- Focus attention where it matters most: Not every AI initiative carries the same level of risk. The principle of manage by exception allows organizations to focus governance and oversight on areas with the greatest business impact or organizational risk, while minimizing bureaucracy for lower-risk initiatives.

- Focus on products: Metrics such as the number of AI licenses deployed or prompts generated reveal little about whether an initiative has delivered value. Success is measured by improvements to business performance, stronger customer outcomes and demonstrable business benefits.

- Tailor governance to the context: Every AI initiative is different. A low-risk internal productivity tool does not require the same governance as an AI application that influences customer decisions. PRINCE2 Agile’s principle of tailor to suit the project encourages organizations to apply governance that reflects the specific level of risk and complexity involved in a project.

Building capability

One lesson has consistently emerged from conversations with organizations beginning their AI journey: technology alone does not deliver transformation. People need to understand where AI can genuinely improve the way they work, how processes can be redesigned around those capabilities and where human judgement remains essential.

We describe this as becoming AI native: developing the habit of asking, whenever a new task arrives, "How can my AI assistant help me with this?" Building that habit helps teams identify worthwhile opportunities before deciding which technologies they need.