AI Projects: Ensuring Success

Insights shared from big names like Vodafone, Mashreq, and Kingfisher; who discussed how they ensure successful AI projects.

Embarking on a new AI project can be rewarding, but daunting. During the December 2024 Financial Times Global Boardroom event, we heard about the latest AI developments within companies like Vodafone, Mashreq, and Kingfisher. They gave insight into the projects and what steps have been taken to ensure success. Adequate preparation and planning is crucial for project ROI, especially with AI. As that worrying Gartner stat goes, 85% of AI projects fail – but avoiding that failure is possible. Read on to find out how.

Use Case Selection

Something that all three company reps spoke on was the process of selecting which use case to pursue. Vodafone CEO Margherita Della Valle said she focusses on impact, not number of use cases, and this philosophy drives her AI approach. AI is seemingly limitless, meaning hundreds of different ideas for use cases crop up in the early stages of project planning. Della Valle expressed that because of this, it is crucial for projects to be focussed. Vodafone rigorously examines each use case and selects a few good ones to explore completely. They purposefully do not run lots of experiments simultaneously. Although it can be tempting to exhaust all the options for AI implementation in businesses, it often leads to frustration and project failure.

Fidra Analytics wholly advocates for this approach. Too often we see perfectly good projects sink from unclear visions and unnecessary distraction. Any AI project we embark upon has been subjected to that necessary scrutiny. We iron out the project’s purpose, expectations, and what “done” looks like. We therefore avoid much of the distraction that Della Valle warns of in the panel.

Good AI Governance within AI Projects

Mashreq Group CEO, Ahmed Abdelaal, raised another crucial point. He emphasised Mashreq’s comprehensive framework that governs everything around their AI ventures. They ensure any AI tool in use never makes critical decisions alone, through a “human in the loop model”. To combat hallucinations and ethical concerns of bias, Mashreq audit systems regularly. They analyse both the data going in to an AI model, and the data it then generates, to ensure the tools stay performing as expected. By prioritising good AI governance, Mashreq have both mitigated risk and ensured that any company AI adoption is smooth. Teams are in the loop and know how to assess the quality of their AI, which gets the tools better quicker.

This approach is an excellent example of how good governance guides AI adoption, not slows it down. AI projects do require frequent audit and assessment. The tool’s ultimate success depends entirely on quality of output and how well it meets the needs of the team. Without regular check ins, this is impossible to ensure.

How does Fidra Analytics ensure success?

Each company raised important points during the panel. We found it rang true for much of our own experience in this industry. We specifically designed our Accelerator Programmes to encompass all these steps – from rigorous use case selection to continual evaluation. If you would like to learn more, we write case studies on our largest projects and post them on our socials.

If you have any questions for our team, email us at info@fidra.ai

www.fidra.ai

Back to Insights