Introducing AI: Focus on what makes business sense

Before you start deploying AI, it is important to identify and understand your use cases. You want to understand where AI could really make a difference and the pain points and challenges that may come with AI deployment.  

Clients that we work with are doing this at both an enterprise-level – AI that will be rolled-out across the organisation, such as AI integrations into everyday enterprise tools like email and word processing applications – and also at a functional level, focussing on how individual departments or specific functions within the business may deploy more specialist AI tools to help with specific activities, such as AI-driven marketing, finance or supply-chain platforms. 

Once use cases are identified, we can start to think about what actions need to be taken to successfully deploy AI, for example, which applications would be low risk, and which would be higher risk, and which deployments would carry regulatory or other implications. 

In this way, identifying and understanding use cases is a foundational step for building good governance around the deployment of AI. 

Customers increasingly expect their suppliers to harness AI to drive innovation, boost efficiency, and deliver competitive advantage. But just as AI itself may present some new challenges, procuring AI can add a few extra twists to negotiations with potential suppliers.  

Strong supply chain management is central to operational resilience and effective risk management. As AI becomes more embedded in business processes, it brings new opportunities - but also new risks - that need to be addressed thoughtfully. 

That means businesses need to approach the process in a transparent, collaborative manner, and establish a clear understanding of the AI system being procured, what data powers it, how it works, and what risks or benefits it might bring.   

Crucially, the intended use of an AI system matters. For example, if AI is deployed as a safety component within critical infrastructure, it could be classified as "high-risk" under the EU AI Act - triggering additional regulatory obligations. That’s why thorough due diligence is essential. 

With a solid understanding of the regulatory environment, businesses can properly assess how risk should be allocated and reflect that in their contractual arrangements. 

That same insight should also inform the design of policies, controls, resilience measures, monitoring and governance frameworks - both for suppliers and customers - ensuring AI is deployed in a compliant and responsible way.  

We’re currently helping clients from a variety of sectors to: 

• review and update their standard contract terms and negotiate terms with suppliers to reflect AI-specific risks, and 
• build robust governance frameworks that support the procurement, deployment and use of AI in a way that’s not only safe and risk aware but also commercially effective.  

In today’s evolving regulatory environment, having the right procurement strategy is key to making AI work for your business. 

Some organisations we’ve been working with are at the sophisticated end of AI procurement and AI governance, and have successfully built AI into key processes and work flows. 

For those that haven’t just yet, there’s often something to be said for taking the easy wins. However wary you are about AI, there are steps you can take to make progress in line with your risk appetite. 

You don’t have to put all your eggs in one large basket. You can start with a few small programmes – easier to set up, control and evaluate. 

You can even try some of the AI applications that are available for free – as long as you’re careful about the information you put into them. 

Pick the low hanging fruit – the ring-fenced content creation system in a team that wants to try new approaches, or the stand-alone analytical application you can just pull the plug on. 

If you’re more ambitious, you could think about volume applications where AI completes structured and repetitive tasks at light speed compared to human workers. 

Providing you have appropriate guardrails, the business case for applications like these is usually pretty unassailable.