Of course, with such innovations comes the fear of human redundancy, but the truth is that technological advancements actually create new opportunities, new jobs, and the need for new skills. It’s not about redundancy, but about retraining or reallocating talent.

We believe in using generative AI to empower people, not replace them. It starts with looking at what tasks are the most tedious or repetitive. The creative or high-value stuff that employees love to do? Generative AI is not going to touch that. Instead, you’re going to automate those low-value tasks, giving employees more time to do what they love, and enabling them to do it better.

Plan, Prepare, and Pilot

Many generative AI projects we see are targeted to start fresh, reinventing roles and overhauling processes. That’s where businesses stumble. They try to accomplish too much too quickly, often with the wrong goals in mind.

We saw the same pattern with the initial hype over cloud and mobile. With every tech revolution, there are those who view technology as a magic wand to instantly fix all issues—a mindset that has never proved fruitful. We live in a practical world where serious business problems demand serious business solutions, not magic wands and wishes.

That’s why we talk about the foundations and applications of generative AI, rather than just the applications. You need the first to succeed with the latter.

The wise move is to tackle your immediate issues, align the technology with processes, and focus on localized generative AI applications. Small-scale prototypes can be developed quickly to help your team understand the technology and provide concrete data for ROI calculations. Achieving value in a few weeks or months is viable, as long as you adjust your expectations and start with smaller, more immediate impacts.

Plug-and-Play and DIY? Think Again

A common mistake is rushing to develop an in-house solution without proper understanding or preparation. Racing to be the first to market can be exciting, but generative AI isn’t plug-and-play technology by any means, and the do-it-yourself approach is about as successful with generative AI as it is with brain surgery. The complexity of the technology and the lack of required know-how make such races futile, often leading to results that are, at best, hit or miss.

We saw this problem arise with one of our clients. Wanting to be one of the first companies to launch a generative AI solution, its internal teams spent a year building a knowledgebase for customers and advisors. However, the knowledgebase was hampered by accuracy errors that are unacceptable in a highly regulated industry.

Thinking it was a content problem, the client approached us about rewriting and restructuring the content, performing prompt engineering, and developing a performance dashboard. To evaluate the content’s performance versus the technology’s, we built our own version of the AI model in about three weeks, applied our recommended settings, and loaded in the restructured content. The result was an immediate 30% improvement in accuracy, leading the client to take our learnings and apply them more broadly across the organization.

Navigating the Complexity

Setting up an effective generative AI solution is a complex task, involving data, processes, and technology intricately linked together. There are custom-built generative AI applications available for everything from writing support to image and video generation, but they all operate in silos—perfectly good for very specific use cases, but unable to collaborate.

And that brings us back to the people element. The simple fact is that there’s more need and demand for skillsets than there are skillsets. People say the technology just doesn’t work, but it’s because they don’t have the right qualified people working on it. They’re just not available.