A Leader's Guide to Navigating Generative AI

Are you a technologist? A marketer? An executive? An analyst? Whatever your title, chances are you have a role to play in the adoption of generative AI (GenAI) in your organisation, given the democratised nature of the technology. Bias and trust have been headline discussion topics in this arena, but what other principles should leaders keep in mind when navigating GenAI's opportunity? Though non-exhaustive, this list calls out some of the issues I have witnessed.

Prioritise Use Cases over Technology: Start with clearly defined business problems GenAI is well-positioned to solve, rather than focusing solely on the novelty of the technology. Successful integration begins with understanding how GenAI can provide value to your organisation. For example, in the retail sector, GenAI can enhance customer experience through personalised experiences and editorial content. Focus on delivering tangible outcomes from the start.

Clarify Ownership: To enable these outcomes, GenAI programmes should be owned by commercial and operational leaders – not just technology leaders. Delivering impact with large language models (LLMs) depends more on understanding use cases and integrating them into business processes than on technical capabilities or model tuning. Business ownership here is critical as it ensures AI initiatives are aligned with strategic goals and operational needs.

Prepare for ROI Scrutiny: While we are very much in the hype cycle now, initial board enthusiasm for GenAI initiatives may diminish as ROI questions arise. Similar to the initial excitement around big data, once passions settle be ready to demonstrate measurable benefits and justify investments in GenAI. McKinsey recently highlighted companies need to have a robust framework for measuring AI's ROI if they want to maintain long-term support from stakeholders.

Recognise Challenges and Blockers: GenAI evangelists might make everything seem frictionless, but significant hurdles exist on the path to production. These include data quality, process mapping and integration, business change, and evolving regulatory requirements. For instance, integrating AI into legacy systems can be fraught with difficulties, requiring substantial change management efforts alongside investments in infrastructure and application modernisation.

Most organisations choose to work with partners while developing these GenAI capabilities. This is sensible, given few have sophisticated AI teams in place with pre-existing GenAI expertise. That said, given both the novel and crowded nature of the vendor landscape, there are a number of factors to keep in mind when choosing a partner or off-the-shelf solution for your generative AI needs.

Critically Evaluate Market Solutions: Many startups offer services that are essentially thin wrappers around existing models like ChatGPT or Claude, adding limited additive value. For more robust implementations that create long-term value, consider partners that focus on building unique capabilities alongside GenAI models – such as those enabling querying and retrieval against an organisation's existing knowledge base, particularly through an industry lens.

Demand Industry Use Cases: While promising use cases for GenAI exist (e.g., document summarisation, code assistance, content generation), few have yet been scaled in production. Challenge partners on their industry-specific solutions – making sure they understand relevant blockers. For example, while healthcare has seen successful GenAI applications in diagnostics, widespread adoption remains limited due to regulatory and integration challenges.

Understand Provider Incentives: Recognise that technology providers often aim to drive consumption or sell services rather than deliver business impact. Evaluate solutions critically to ensure they align with your strategic goals, not just vendor offerings. For instance, some AI vendors might prioritise upselling their premium services rather than customising solutions to fit your specific needs.

Balance Insource and Outsource: Once you find a partner, it can be tempting to outsource the majority of projects to them. After all, they have the skills and the resources required. However, like a crutch improving mobility, this approach does not allow the organisation to build the muscle required to be independent. Therefore, carefully consider how and when you build your internal capabilities.

There is no doubt the opportunity presented by generative AI is both real and significant, despite the frothiness of the current market. Yet, like the excitement around big data before, the long-term success and ROI of initiatives will depend as much on leaders' critical thinking as on the technology's potential. Hopefully the principles listed in this post will help tip the odds in readers' favour.

– Ryan

Cover image by DALL·E.