Success requires a clear strategy, a focus on the right business problems, and a commitment to integrating technology in a way that truly transforms how you work.

While many companies are placing big bets on AI, a recent report from MIT’s NANDA initiative, The GenAI Divide: State of AI in Business 2025, reveals that most of these ventures are failing. According to the report, a staggering 95% of generative AI pilots are stalling, delivering little to no measurable business impact.

Despite the widespread enthusiasm, there’s a clear divide between the few success stories and the many stalled projects. The research, based on interviews, surveys, and analysis of hundreds of AI deployments, highlights a crucial point: the problem isn’t with the AI models themselves, but with how companies are implementing them.

The Real Reason for Failure

The report points to a “learning gap” within organisations. While individual users get a lot of value from flexible, general-purpose tools like ChatGPT, these same tools often fail in an enterprise setting because they don’t adapt to specific workflows.

The report also found a significant misalignment in resource allocation. More than half of all generative AI budgets are going to sales and marketing tools, yet the highest return on investment is found in back-office automation—things like cutting costs on external agencies, eliminating outsourcing, and streamlining operations.

What Makes an AI Project Succeed?

The research offers some clear insights into what makes a successful AI deployment:

• Partnerships over Solo Projects: When companies partner with specialised vendors, their projects succeed about 67% of the time. In contrast, internal builds succeed only about 33% of the time. This is especially relevant for highly regulated sectors, where many companies try to build their own systems but often fall short.
• Empower Line Managers: Successful companies empower their line managers, not just central AI labs, to drive the adoption of new tools. This ensures the tools are relevant and used by the people who need them most.
• Integrated Tools: Success comes from selecting tools that can integrate deeply into existing systems and adapt over time.

Workforce changes are also underway. The report notes that companies are not backfilling positions in roles like customer support and administration as they become vacant, rather than conducting mass layoffs. These changes are primarily affecting roles that were previously outsourced.

The promise of AI is clear, but so are the challenges. Simply buying a generic AI tool and hoping for the best is a strategy that the data shows is highly likely to fail. At CREAPLUS, our team of AI and cybersecurity experts can help you develop the right strategy and guide you through the implementation process. From selecting the right tools to ensuring a secure and effective rollout, we can help you avoid the common pitfalls and build a solution that delivers real, measurable value for your business.