The approach many enterprises are taking to implement AI is akin to the search for the Holy Grail. Yet, for all this ambition, very little is actually going live. According to McKinsey 90% of AI pilots fail to make it into full production, suggesting there is a growing disconnect between boardroom excitement and real-world execution.
This inertia is due to several factors. Implementing AI is a complex process and requires organisations to navigate a minefield of domains. Whether its managing compliance, assessing technical capability, enhancing integration, or proving the return on investment (ROI), each step demands considerable resources and inter-departmental collaboration.
The fixation on big, transformational AI initiatives is also contributing to this lack of progress. These are the moonshot AI projects with the potential to deliver massive productivity gains and distinct competitive advantages. It’s a tantalising proposition. Get it right and lead the market, but the resources required are often underestimated.
NASA calculated that it took the combined efforts of 400,000 people to accomplish the Apollo programme. Moonshots by their very nature require a lot of people pulling in the same direction to get off the ground. This is easier said than done, and organisations that bet everything on big projects can easily find themselves in a doom loop of endless proof-of-concepts. Welcome to Pilot Purgatory: where good ideas go to die. Not because they weren’t viable, but because there was no clear link to the business impact and business ownership of the project. Too often we see technology solution looking for a business problem.
Less can be more
But there is another way. Over the last number of years, I’ve observed dozens of organisations adopt and implement AI technology to varying degrees of success. The businesses that have the most success follow a distinct pattern: they get AI tools and capabilities into the hands of users quickly.
This is ‘micro-AI’ in action. There’s a host of new and often unused AI capabilities already embedded in the business applications that workers use day-in day-out. By simply switching these features on, they can deliver value from the outset and mitigate the risks and challenges of building bespoke AI tools. It also means there’s virtually no learning curve — onboarding employees onto new systems is one of the biggest (and most overlooked) costs of transformation.
The respective productivity gains of adopting individual micro-AI features can be relatively modest. On their own, these are small changes — each tool may only improve efficiency by 1-2%. Together, they can be truly game-changing. When combined across a typical user’s workflow, significant productivity improvements of more than 20% are realistically achievable.
One way to think of this is having a team of 100 interns ready to support every function of your business. By freeing up time and reducing friction across multiple processes, workers can dedicate more time to the high value work that impacts the bottom line. Each AI ‘intern’ can make a small, but measurable contribution at every level of an organisation, helping everyone else move faster and get more done.
A great example of this in action is The Very Group, one of the UK’s largest online retailers. The company activated AI goal-setting features in Oracle Fusion Cloud Human Capital Management (HCM) for 2,500 employees with the switch of a simple toggle. The AI capabilities help managers create SMART goals and assist employees with personalised suggestions, overcoming “blank screen syndrome.” Since activation, the feature has been used 10,000 times, transforming how the company manages performance objectives with minimal effort and zero additional cost.
Watch your step
Regardless of whether you are a successful adopter of AI or you have yet to make real headway, every organisation is at risk of traps that slow progress. Many fall at the first hurdle by waiting for a “perfect” strategy before taking action. Another blocker to progress is the misconception that implementing AI will require significant investment or mean embarking on a complex IT build. This is despite micro-AI benefits being easily accessible to almost everyone.
At the other end of the spectrum, fast-moving businesses may bite off more than they can chew and spread resources thinly across too many initiatives. Conversely, large organisations are sometimes guilty of wasting resources by deploying AI in silos with no cross-functional roadmap or consistent group-level vision. There are very few organisations getting everything right and there is still room for all of us to improve.
The key insight is that while transformational AI projects capture the imagination, the path to AI success often lies in the accumulation of smaller, more manageable improvements that are driving a clear business metric. These micro-AI implementations provide immediate value, reduce risk, and create a foundation for more ambitious projects down the line. The organisations winning today understand that sometimes the most powerful strategy is the one that gets you moving, not the one that promises to move mountains.
