Blog by Josef Novak, Chief Innovation Officer, Spitch

The Generative AI market is caught between two extremes: endless hype and reality. On one hand, there is clearly incredible potential. On the other, a report from MIT’s NANDA lab earlier this year suggests that despite $30-$40 billion in enterprise investment, almost 95% of organizations saw zero return on their recent AI initiatives. This sounds shocking but most of this comes down to poorly organized workflows or misalignment with actual business goals.

This isn’t really an indictment of the technology itself. It’s a reflection of a predictable cycle of overenthusiasm. The critical challenge today is not that different from any other point in time or other new technology: it is learning how and where to adopt that new technology. It requires adopting GenAI technologies selectively, for use cases where they can actually make a positive business impact.

Key Takeaways

• Agentic AI with human-in-the-loop implementations should aim to improve overall organizational excellence by empowering individuals across the organization.
• Pilot projects must be designed to solve a specific, tangible business problem.
• Rather than attempting to overhaul everything, focus on intelligent automation that augments existing processes.

The Skills Gap is a Strategy Gap

An interesting question is whether this AI skills gap underlying these unrealized returns is a local or global phenomenon. The answer however, seems clear: the challenges are universal. Around the world, technology is evolving at a pace that makes it incredibly difficult for businesses – and even experts – to keep up, both in terms of finding talent and adapting business processes.

Expertise levels and appetite for change vary by culture and organization, however the fundamental challenge remains the same. We are all navigating a new technology landscape and, in many ways, still figuring it out together. The most successful organizations aren’t just hiring data scientists and machine learning experts; they’re bridging this uncertainty gap by building a strategic framework for adoption – which carefully considers use case appropriateness and not just enthusiasm for new tech.

This framework begins by focusing on people and the organizations they make up. It involves educating leadership and teams on AI’s real-world potential and just as critically its current limitations. From there, the key is to start small, but think big. By co-creating pilot projects that solve a specific, tangible business problem – like advanced agent training – organizations can provide a hands-on learning experience, demonstrate value quickly, and build momentum for broader change. They can also conduct deeper analysis of these use cases as they evolve. This approach should be guided by a strategic roadmap that aligns AI initiatives with core business goals, ensuring a clear integration path for both new and existing technologies.

Finding the Quick Wins Among the Challenges

There is currently intense pressure to deliver quick wins on AI investments. However, as mentioned earlier, the most common challenges have little to do with AI models themselves and more to do with standard business challenges:

• Access to data that is often siloed or difficult to reach.
• Unclear problem definitions or poorly chosen targets.
• Integration issues with legacy systems.
• Cultural and/or organizational resistance to change.

The most effective way to generate immediate value is to identify opportunities that cleverly subvert or short-circuit these problems.

This might mean using AI services to predict and pre-complete an agent’s post-call tasks, or providing one-on-one AI training materials that are tailored to the needs of individual employees. The MIT NANDA paper was critical of many projects, but it also identified a key area where Gen AI consistently shines: use cases that increase individual productivity.

This is where we should focus our energy. Tools such as Gen AI-supported journey editors or in-call agent assistance directly leverage this strength as well as the models’ core capabilities. The goal then is to improve overall organizational excellence by uplifting the individuals within it. In this setting, AI plays a powerful supporting role, but the human remains firmly in the loop, in charge, and responsible. Analyzing the outcomes of these focused initiatives helps us understand why we succeed and where to focus our energy next.