Over the past few weeks, Fabian Haijenga and Koen van Nistelrooij have held over 20 conversations, among them with CxOs, about the potential of AI. Their conclusion? The potential is enormous, but the real strength lies in integrating AI within the unique context of an organisation.
Apple's recent presentation of their AI vision at WWDC24—Apple Intelligence—underscores the importance of contextualised AI. For example, why would a lawyer [or insert your expertise here] set aside their 30-year archive of experience (read your SharePoint/Drive with thousands of documents) and rely on generically developed AI models? Imagine if your organisation could provide all its thousands of hours of accumulated expertise as context (or training) for the AI? Context is crucial; without it, AI remains generic, prone to errors (hallucinations), and lacks the expertise of the specific business environment.
Moreover, we should not limit AI to flashy innovations (“shiny objects for the outside world”). True integration within an organisation requires a deeper, more holistic approach.
Historical trends show us that initial enthusiasm often fades, resulting in missed potential value and, worse, growing skepticism. We’ve seen this with business intelligence, big data, block chain and robotisation. These initiatives started with high ambitions but often failed to achieve fundamental change, let alone adoption.
Why does this happen?
- Superficial approach: Projects start small and simple instead of small and essential. They look flashy but don't solve core problems and are predominately focused on IT
- Expert bias: AI is (already) a broad field requiring multiple expertise. This leads to applications being sought from the perspective of an expert (a carpenter prefers to work with wood), rather than a holistic approach where multiple solutions are to be considered. Using AI should not be a goal in itself
- Limited adoption: Usage usually remains limited to a small group because users were not involved in the development process, let alone included in the rollout, leading to solutions that don't quite fit their needs and aren't used
- Isolated solutions: Good initiatives are developed in isolation and not connected to other initiatives, solutions, and procedures. This doesn't mean that every initiative needs to be grand and entirely overhauled, but it must be structured and purposefully aligned with broader goals.
Tips for your AI strategy and adoption
- Stay informed and inspired: The AI landscape is changing rapidly. Regularly update yourself and your teams with the latest developments and trends. This should be an ongoing activity, not a one-time workshop
- Identify opportunities (use cases): Systematically look for areas within your organisation where AI can have a significant impact. Structure this process identifying pain points and opportunities, and visualising where AI can introduce efficiencies or innovations. Make it smart including the potential value, requirements (including required context data) and implications
- Quantify potential: Assess the feasibility, urgency, and financial value of potential AI initiatives. Prioritise projects based on their expected return on investment and strategic importance. Use KPIs to clearly measure the potential impact. Be bold in driving whether or not these KPIs are met—‘Kill your darlings fast’ in all phases of development
- Start with proof of concepts: Before scaling up, implement small, targeted proof-of-concept projects (“AI Living Labs”). These should be designed to build and validate ideas within a short period of 12 weeks. This approach allows for quick iteration, learning from failures, and refining successful strategies. Ensure each proof of concept has clear objectives, measurable outcomes, and a path for scaling up if successful.
- Encourage co-creation in mixed teams: Form a cross-functional team and assign clear ownership. Ensure all stakeholders, including end-users, process owners, and experts from security & privacy, are involved from the start. This helps avoid unnecessary complexity and ensures the project meets actual needs and operational realities
- Energetic icebreaker: Ensure you have a captain who, with great energy, acts as an icebreaker to get initiatives moving, brings together colleagues of varying disciplines to work together, pushes through when things get tough, and creates a wave that others are eager to ride
- Ensure data quality: Not the most sexy topic, but good AI depends on good data. Garbage in is garbage out. Make data quality a requirement by understanding your data landscape, effectively cleaning and structuring data, and maintaining data governance practices
How to start
As a decision-maker, it’s crucial to take the lead in setting expectations and direction for your AI initiatives. Start by clearly defining your AI goals and aligning them with your organisations strategic objectives. Communicate this vision at all levels, foster a culture that supports innovation, and ensure you have the necessary resources and stakeholder engagement.
At NimbleNova, we have an awesome team ready to help you on this journey, from initial strategy development to the successful implementation of AI labs and further scaling. Feel free to DM/App/call Koen van Nistelrooij or Fabian Haijenga to discover how we can accelerate your AI ambitions.
Not ready for a conversation but want to gather your AI initiatives in one place? Type "AI-initiatives" in the response below, and we'll send you a template we use in our workshops.
PS. DALL-E did not agree with us about whether this is the front or the back of the icebreaker 😎
Share this blog: