08/09/2025

Only Five Percent Succeed with AI

Only Five Percent Succeed with AI

How to Avoid Being Among the Losers

Lessons from MIT’s Report

MIT’s NANDA initiative has analyzed over 300 AI implementations. The conclusion: successful companies share a common approach – they choose a clear pain point, execute quickly, and build partnerships instead of trying to develop everything in-house.

Those who fail often invest heavily in their own internal solutions that never truly become part of everyday operations.

“We often see companies building too big and focusing too much on the technology. AI works best when it’s woven into workflows and solves a concrete problem,” says Perslow.

From Hype to Real Value

At HiQ, Sofie Perslow and her team work to break this trend. The key is to start small, but with the right structure.

“We often tell our clients: choose an area that really matters. A bottleneck in customer service, a heavy process in back office, or a recurring error that creates rework. If you apply AI there, the value becomes visible right away,” she says.

MIT’s report points to the same pattern: the biggest returns are not found in sales and marketing – where much of today’s AI budgets are spent – but in back office and administrative processes that are repetitive and resource-intensive.

“It may not be as glamorous as building new automated sales tools or generating marketing material, but this is where companies can first and foremost free up resources and strengthen competitiveness in a real way,” says Sofie.

How to Become One of the Five Percent

What separates the winners from the rest? According to Sofie, it comes down to five things:

  1. Focus on what you want to achieve, then choose technology, method, and tools accordingly – measure and follow up.
  2. Integrate AI into workflows; AI that’s used daily makes a difference, not experiments in the lab.
  3. Build partnerships; invest in proven solutions and bring in experts rather than reinventing the wheel.
  4. Do the groundwork properly. Without good, accessible data and clear processes, success is unlikely.
  5. Start small and test where you can create value, then scale up to make a real impact.

“With this approach, we see that even large organizations can succeed in AI transformation and increase their chances of being among the winners,” says Sofie.

The Next Wave: Agentic AI

Already today, the most advanced organizations are testing the next generation of AI – so-called agentic AI – where systems can remember, learn, and act independently within defined parameters.

Finally, Sofie Perslow warns against being discouraged by the report’s findings.

“Waiting or abstaining from implementing AI where it can make a difference is not an option if you want to build a sustainable business going forward. As with all major shifts and changes, it takes time to learn – but it’s only by trying and testing that the real lessons can come.”

Contact

Region

You want to know more? Let’s get in touch!