Accelerating AI Adoption: A Competition for Best Use Case at Discovery Limited
Accelerating AI Adoption: A Competition for Best Use Case at Discovery Limited
Summary
This case study explores Discovery Limited’s AI adoption journey. The South African financial services organization got ahead of AI-adoption resistance by launching a PMO-led campaign with free PMI courses, a prompt library and a use-case catalog, culminating in a competition for success-story sharing. As a result, shared use cases grew from a few to 45, reflecting stronger engagement and experimentation across the PMO community.
Key Takeaways
Here are some of the key takeaways from Discovery’s AI journey:
AI adoption depends on literacy, trust and practical support — not just access to tools. Discovery found that employees needed training, clear governance, approved tools and real examples to feel confident using AI in their daily work.
The project management office (PMO)-driven campaign and competition helped shift AI from leadership push to employee pull. By encouraging learning, success-story sharing and friendly competition, Discovery made AI experimentation feel more approachable, visible and motivating.
The winning use case aligned with Discovery’s strategic goal of demonstrating AI’s productivity value while keeping people in the loop. The winning user acceptance testing (UAT) example showed how Microsoft Copilot helped reduce weeks of manual test-case creation to just two days, while reinforcing that human judgment remains essential.
Challenge: Scaling AI Adoption
Discovery needed to scale AI adoption across a large, federated organization, where teams worked on projects in different product areas, and had uneven, often insufficient, levels of AI literacy. Discovery’s GenAI Academy recognized that simply providing tools like Copilot wasn’t enough; people needed shared governance, responsible-use guidelines, approved tools, training and a common understanding of what AI could do. Adoption was slow at first because some employees saw AI use as “cheating,” others were too busy to experiment and overall literacy levels varied widely. Discovery had to build trust, establish guardrails, create communities of practice and give people practical, work-related use cases so AI adoption could move from small pockets of interest to broader organizational value.
According to Jonathan Zondo, head of PMO — Group Information Services (GIS), which is responsible for providing technical services for the organization, “The real challenge wasn’t deploying AI — it was helping people understand it, trust it, find time to learn it and see how it fit into their everyday work. Once literacy improved, adoption followed.” He described the challenge as a multistage change management effort rather than a technical rollout.

Approach: The “Curious PMO Conversations With AI” Campaign
The turning point came when Discovery focused on raising AI literacy across the PMO community. Zondo emphasized that when employees had common training, access to examples, a prompt library and opportunities to share ideas, adoption shifted from being mostly “push” (driven by leadership) to increasingly “pull” (employees actively seeking more AI capabilities).
This approach enabled Discovery to overcome several challenges to adoption:
Scaling AI across a federated organization
Discovery consists of multiple autonomous business units, supported by nine PMOs across the South Africa composite, with different ways of working.
The organization provided the necessary tools and allowed people to evolve at their own pace, since individual groups interpreted use cases differently and had varying levels of maturity and needs. Zondo said this led to the creation of a “hub-and-spoke model: The Gen AI Academy provides the curated material training. They provide the structure and the strategy in terms of where the organization is going. Each one of the professional disciplines then has their own focused AI community, like we have in our PMO. And each one is enabled to use our internal tools, experiment, surface what they’ve experimented and then realize that good value going forward.”
Addressing initial skepticism and fear
Many employees viewed AI as “cheating” rather than a legitimate productivity tool.
“We needed to go on an education drive first to inform people that if you are not using AI, you are falling behind,” said Zondo. “The next evolution of knowledge workers is not going to be based on their level of qualifications and education and experience. It’s going to be based on how they can use AI to augment their day-to-day capabilities, to provide a more visionary role that sets them as consultants, in advisory roles. AI is going to enable us to take away all the coordination. We’re going to use it for orchestration, and the human being is going to then be that difference-maker.”
Building trust through governance
Before broad AI adoption, Discovery established responsible-use policies, security guardrails and a “human-in-the-loop” principle.
Zondo emphasized that leadership needed to reassure employees that AI was intended to augment people, not replace them. “We were very deliberate about ensuring that people must experiment within our work frames. Look within how we currently work, within the frameworks that we have in place, the process that we have in place, and identify those areas that could benefit from the use of AI.”
Overcoming slow adoption
Literacy was the major barrier. Even after Discovery established an AI community and provided approved tools, adoption remained limited and concentrated in small pockets. People knew tools such as Copilot existed, but many did not know how to use them effectively or keep up with new capabilities.
Discovery PMO AI Community launched the “Curious PMO Conversations With AI” initiative to help people “cross the chasm between the crazies who will adopt any technology and the early majority,” said Jaco Viljoen, generative AI (GenAI)/agile coach at GIS PMO. The effort “was less about teaching AI technology and more about helping project managers take small first steps, learn from peers and see concrete examples of value creation so adoption could spread organically.” The use of the word “conversations” was “to show that it’s for curious people,” said Viljoen. He explained that the goal was to foster conversations about AI and its value: “That was the idea behind the campaign.”
Finding time to learn new skills amid competing priorities
The team worked with managers to create development opportunities during work hours rather than expecting people to learn on their own time.
"After doing the courses," Viljoen said, "we urged people to experiment with AI. If you don't know what to do, use those use cases [found in the PMI Talking to AI course] and apply them in your work environment. Make things happen, and then once you've done that, write a success story."
Solution: Sharing Success Through a Competition
A major component of the campaign was helping project managers document AI successes. Recognizing this need, Viljoen created a framework and a series of prompts that guided participants through writing success stories, including quantitative metrics where possible. To generate excitement, the campaign culminated in a competition.
Viljoen said he realized from working with people that writing success stories is not easy. “The mindset is the challenge. People just don’t think about success. It works, OK, great, let’s move on. But what was that success? I created a framework that leads you through a series of questions, and you just answer the questions. And then we would prompt them a little bit to expand on it. I would use AI together with some guidelines that I’ve created specific for AI on how to write a success story.”
The final step was encouraging people to share their stories. Discovery set up an event, primarily in person with virtual participation. “The idea was to create a little bit of excitement about it, get people to engage with it, have a bit of fun, win something,” said Viljoen. “Each one of the PMOs had an opportunity to submit one to three of their success stories. We put together a panel of judges and awarded a winner — first, second, third place with prizes.”
Presenting a Winning Use Case
Chantelle Smit, test manager, UAT, found herself unexpectedly entered into the competition. Managing a team of testers, Smit faced a bottleneck in the labor-intensive process of creating UAT test cases. Testers could spend up to six weeks manually converting business requirements into test cases, delaying testing and creating resource constraints across the team.
Smit thought leveraging an AI-powered assistant might help. “My first thing was, there must be a way to write these cases quicker, faster, easier. I didn’t know if it was going to work, to be very honest. I just took a wild gamble.” Smit was able to reduce the effort from weeks to days, allowing testing to begin much sooner. “Something that probably would have taken us another six weeks plus, it took me two days to write almost 400 test cases with Copilot. AI for us as UAT is almost completely eliminating all the manual effort.”
Smit’s manager entered her into the competition. “I’m not a public speaker. I have an absolute phobia for it,” Smit said. “He put my hand up for me in a previous meeting to say that I will present. There were about eight or nine project managers that were competing. I was the only person there from testing.”
Smit’s use case encompassed Discovery’s view of the future of work: encouraging innovation, enhancing productivity and keeping the human in the loop. “It doesn’t take away anybody’s work or what anybody does. It just alleviates some of the manual effort. That human intervention is still so important because you do need to ask it properly for it to give you what you want,” said Smit. “AI is absolutely so powerful. It opens so many doors for you personally. If you can’t see the benefit of it opening doors for you and your work, you’re going to be left behind.”
Building a Repository for Propelling Future Value
“The value that we as Discovery got are those success stories,” said Jaco Viljoen. “I created a central repository that is available to everybody. So, if I’m a new project manager coming to Discovery and I say, ‘Everybody’s using AI; how can I do it?’ Take them to the repository of success stories.”
Chantelle Smit said the campaign prompted her team to keep experimenting with Copilot, leading to tangible improvements in how they design workflows to reduce manual effort and save time:
- Cut test-case creation from weeks to days
Copilot reduced a six-week manual process to about two days.
- Generated nearly 400 test cases
Copilot converted business requirements into structured test cases, including Cucumber-format outputs for automation.
- Removed a major bottleneck
AI shortened weeks of test-case writing so testers could start much sooner — sometimes the Monday after Thursday/Friday walkthroughs.
- Reduced administrative work
AI did much of the requirements-to-test-case drafting, while humans reviewed and refined the output.
- Improved the process
Copilot-generated test cases plus a Jira import script let the team prepare finalized cases for execution in minutes.
- Freed up capacity
Time savings eased resourcing pressure and created room for upskilling and higher-value work.
Smit said her manager’s nomination pushed her out of her comfort zone: “I’m very grateful for the work Jaco and the team have done to bring this in and what they’ve given us, because my team uses it not just for manual testing or creating test cases. I trust it. I love it. I absolutely love it. I use it for everything.”
Conclusion: AI Adoption Moves From Push to Pull
The “Curious PMO Conversations With AI” campaign is helping AI adoption at Discovery move from push to pull among the project management community. “We want more of this,” said Jonathan Zondo. “There’s starting to be that pull, but we haven’t quite measured it. I still don’t think we’ve reached a 50-50 balance. In an ideal world, it would be great if we could have a 70% pull and a 30% push because it means that the message has landed. There’s a common shared vision and a realization that if we all work together, there’s greater value that we can achieve. I think we are getting to that stage.”
Zondo said the next phase is to take the value achieved in the campaign and share it more widely. “We’ve been able to create a shared space where all the new prompts and use cases reside. And we are starting to unearth additional cases with PMI Infinity™. When we spoke about PMI Infinity during our campaign, everybody got excited. That created an even greater pull effect because now people want to get involved and understand the power and value PMI Infinity brings. As a result, we are getting more people who want to become Project Management Professional (PMP)® certification holders, which also increases our legitimacy as a project management community.”
Related Content
The Project Professional’s GenAI Journey
Learn how project professionals are using generative AI to drive smarter decisions, improve efficiency, and lead strategic transformation.
Talking to the Machine: Prompt Engineering Essentials for Project Professionals
Prompt engineering is predicated on the interaction between human language and computational power, blending human intellect, creativity and the capabilities of artificial intelligence.
Enabling Project Management Transformation With GenAI
Organizations can unlock the true value of generative AI by going beyond tools and pilots to establish the support structures, resources and culture required for success.
You May Also Like
Certification
PMI Certified Professional in Managing AI (PMI-CPMAI)™ Bundle
No experience required
Invigorate your career with the gold standard certification for leading AI projects and driving real business impact.
