AITalk: Shaping your organisation's AI strategy to solve real-world challenges

On 16 October 2025, AITalk host Kevin Craine was joined by Dinesh Mohan, Global Head for Digital & Technology, Expleo; Nabeel Nawaz, Global CIO M&A Leader, IBM; Victoria Grech, Founder & CEO, Trustentia; and Pawan Udernani, Vice President, Data-Tech-AI Leader, Genpact. 

 

Views on news   

The current surge in AI adoption reflects a rat race of hype-fuelled, entrepreneurial deployments with many businesses adopting rapid AI implementations aimed at achieving quick tactical results and productivity improvements, often without sufficient attention to long-term strategic considerations or overall business transformation goals.

 

A recent MIT study found that 95% of GenAI pilot projects failed to deliver measurable return on investment, underscoring how fragmented, tactical AI applications often fall short of enterprise impact. 

 

Ultimately, however, the impact of AI must be measured by business outcomes, such as revenue growth, speed to market or reduced operational risk – not tech-centric key performance indicators. Companies who have moved beyond pilots may be still in the middle of their AI projects and therefore we may have to wait a few more months to hear about a wave of success stories.  

 

Where can AI deliver the greatest RoI? 

While pilots tend to be successful, only 15 per cent of them scale across the enterprise. A successful journey starts with finding the right proof of concept. The use case must certainly have an impact on the business’s P&L and the right data must be available for running it.  It’s equally important that the aim of leadership should be to enable the human workforce with AI and not replace it. Finding the right talent and engaging them in experimenting with the technology even if they don’t have the right qualifications are essential too.

 

To get the sponsorship of the leadership, those driving change must demonstrate the value of these investments and create a vision that can win the C-suite over. To achieve workforce readiness, organisations need multi-level skills and capabilities. Ay the foundation, the workforce must have the right data skills. Then comes the AI layer followed by the applications and the AI governance layers. The roadmap that companies follow on their AI deployment journey must also integrate iterative improvement. In companies that succeed the common thread is a start-up culture that’s become part of their DNA.  

 

As for people, you don’t need to turn everyone into a data scientist but imbue the organisation with a certain data knowledge and awareness.  While the magic happens when the model gets fine-tuned, data management and governance are the pieces that can lead to the failure of the project. Having champions within each function is key, but it’s also important that everyone in the leadership team speaks the same language and goes in the same direction.

 

CEOs can have office hours where they talk about how they and their teams use AI tools. One of the mistakes businesses make in this area is that they scale pilots, not platforms. Make sure that you have a single governed data plane. You’ll need an orchestration layer too for assistants and agents.  

 

The panel’s advice 

• When pitching an AI investment to leadership, start with the why and then move onto the what and the how. Learn from start-ups. 
• Find the right AI tools for the right issues.  
• Don’t create your own foundation model, they cost a lot of money. 
• The biggest RoI for next year is probably having your own IP and your own open-source model.