Why 95% of GenAI initiatives “fail”... and what that really means

Organisations everywhere are experimenting with chatbots, copilots and intelligent assistants. Yet the latest MIT NANDA report highlights a sobering statistic: 95% of GenAI initiatives fail to deliver measurable return on investment within six months. 

At first glance, this seems to reinforce the idea that GenAI is little more than hype. But a closer look at the report reveals a more nuanced picture. “Failure” in this context does not mean the technology doesn’t work. Rather, it means that within the six-month window following a pilot, no clear profit and loss impact was recorded. For complex transformations, that is an exceptionally short horizon. 

This reflects a reality many leaders already recognise: AI is not a quick win. Success requires more than a proof of concept or a flashy demo that dazzles the boardroom. The real challenge lies in integration and adoption—embedding AI into everyday processes, enabling it to learn from context, and ensuring it delivers tangible business outcomes. 

Why do so many initiatives fall short? 

The reasons are all too familiar. Many projects get stuck in what’s often called “pilot purgatory”: a demo that works in isolation but never evolves into a scalable solution. 

A key factor is the learning gap. Most AI systems don’t retain context across sessions, fail to learn systematically from feedback, and struggle to adapt to real workflows. What impresses in a demo often delivers little lasting value in daily operations. On top of that, many initiatives lack a clear connection to strategic objectives and KPIs. The result: frustration, wasted investment and a growing gap between promise and reality. 

Budget allocation compounds the issue. More than half of AI spending goes to sales and marketing, where results can be showcased quickly on dashboards. Yet, as MIT highlights, the less glamorous back-office functions like finance, document management and operations, often provide the most substantial and sustainable ROI. 

What the successful 5% do differently

The organisations that succeed in creating value with GenAI share various common traits: 

• They begin with clear business objectives, not “technology for technology’s sake.” 

• They focus on seamless integration into existing workflows and systems. 

• They develop applications that learn, adapt and evolve with the organisation. 

• They target high-frequency, measurable tasks such as claims handling or document processing. 

• They embed AI directly into workflows rather than building stand-alone tools. 

• They track impact on KPIs continuously and adjust as needed. 

In short: success does not come from experimentation for the sake of it, but from a deliberate strategy with a clear focus on value.

How AE and SD Worx avoided "Pilot Purgatory"

 

At AE we know that lasting success with GenAI comes from co-creation with real users. The SD Worx AskMe case shows how initial scepticism turned into enthusiastic adoption and why AE is among the 5% of GenAI initiatives that deliver measurable, sustainable value. 

---

→ From sceptics to ambassadors 

From the very beginning, SD Worx employees, the daily users, were closely involved as domain experts in every stage of development. Their positive experience encouraged colleagues to join in, accelerating adoption naturally without the need for heavy, top-down change management. 

Because employees saw their own input reflected in the evolving solution, they quickly became early adopters and ultimately ambassadors. Their positive experience encouraged colleagues to get on board, accelerating adoption naturally without heavy top-down change management. 

→ Measurable adoption and impact 

One of SD Worx’s key business goals was to onboard new employees more quickly and effectively. After just one year in production, AskMe has become an essential part of daily work: 

° More than 100 employees use AskMe actively every day. 

° New colleagues benefit most, with a drastically shortened learning curve. 

° Users receive accurate, natural-language answers linked to more than 100,000 documents. 

° Employees report greater confidence and higher satisfaction in their daily tasks. 

---

By involving users throughout the development process, AE delivered not just a tool but a living, evolving assistant that is embraced by the very people it was designed for. 

 

What does this mean for organisations? 

The MIT report is not a reason to question the potential of GenAI. It is a wake-up call to approach it differently. At AE, we see it every day: organisations that invest in strategy, integration and adoption today secure a competitive edge tomorrow. 

Our teams help organisations define realistic use cases, validate them, and embed them in day-to-day operations. Not with lofty promises, but with measurable impact built step by step. 

Why this places us in the 5% 

This user-driven success model explains why AE is part of the rare group of organisations that turn GenAI pilots into lasting business value: 

• We co-create with users to ensure relevance and trust 

• We integrate AI into daily workflows for seamless adoption 

• We focus from day one on adoption and measurable results 

• We build solutions that continue to learn and improve 

The SD Worx AskMe story demonstrates that with the right approach, clear KPIs, user-centred design and a focus on adoption, GenAI can move from experiment to a driver of business innovation and transformation. 

The headline “95% fail” may sound stark, but the reality is more nuanced. Generative AI is not failing; it is stuck in pilot mode at organisations that never make the leap to integration. The challenge lies not in the models themselves but in the hard work of process change and rigorous evaluation. 

The real question is not whether AI can deliver value, but how organisations prepare themselves to be among the 5% success stories. Organisations with digital ambition take the lead and the time to act is now.