When it comes to generative AI, many companies think of ChatGPT and copilots for automating relatively simple tasks. But the real value lies in solutions that structure data and identify trends. These solutions impact the entire organization. While more and more companies are realizing the value of these types of AI projects, they often don’t get beyond a proof of concept. “Too bad,” says Andrew de la Haije, managing partner at Eraneos.
About 10 to 15 years ago, companies started to realize that they could use data to make a difference. In the last few years, there has been a lot of interest in the AI side of that data spectrum. “It started with predictive modeling. Think about predicting where maintenance is needed ahead of time and optimizing things like staff schedules, asset management, and the Internet of Things. Then we saw a real revolution in data and AI: generative AI. Thanks to ChatGPT, this suddenly became accessible to everyone. It was a real revolution that led to many other innovations.
“It’s not a simple solution, GenAI can improve your entire value chain. That’s where the real value is. That’s where the leaders are going to make the difference.”
Optimal use of resources thanks to AI
This revolution made all companies want to jump on the bandwagon. Because they didn’t want to be late to the party, but also because of the faltering economy and the shortage of good employees. “They were looking for solutions to make the best use of their resources, whether they were resources, people, or other things. And rightly so. There is a huge potential for improvement in the implementation of generative AI in companies when it comes to process optimization and cost reduction.”
Not just for one department
The possibilities of generative AI have certainly been picked up on in recent years, but Andrew says something remarkable is happening here. “Many customers are all too eager to have a proof of concept (PoC). The problem is, it often stops there. Think of a 500-person customer service organization that can’t get the job done and should have twice as many people. We do a PoC that demonstrates that the organization can actually get by with 250 people. If all goes well, a business case is tied to it. Suppose it shows that if you spend twice as much money on the PoC, you will save ten times the cost. You and I would immediately say, ‘Do it!’”
Why does the project get stuck after the PoC? “The business case goes beyond the department. Customer Service can work smarter and more efficiently, but the real value is in one or more other departments. That’s where the politics start. There has to be consultation with those other departments. Who is going to pay for this? Aren’t there better solutions? While there are opportunities, this is exactly how companies limit themselves.”
AI as an IT party
Then there is the technical side. What we often see is that a PoC is initiated by IT. “Then you run into the same problems we had 10 to 15 years ago: the idea that IT has come up with something again. Organizations look at feasibility from a very technical perspective, and the business is less likely to be involved.
With the introduction of methods like DevOps and agile, more disciplines in the organization started to work together. With generative AI, we see history repeating itself, and we are still in the early stages.”
The real power of AI
At the other extreme, companies are rolling out AI solutions from the business side. The danger with this is that different departments will look for their own point solution, when the value of a solution transcends different departments. There are many simple generative AI solutions in the knowledge management space. Examples include copilots that write emails or summarize texts based on prompts. But the real power of AI lies in associating semantic events. “For a tram company, we developed a solution where mechanics speak the details of a repair on their tablet in their own language. The power of AI is not so much in speech-to-text or translating those stories, but mainly in recognizing patterns. Failures that repeat themselves, that can be found or prevented more quickly if you are vigilant. Or take a solution for a telecom company that can determine a customer’s mood based on the language and words they use. Again, this is relevant, cross-domain information. It’s valuable for customer service, but the return for marketing, maintenance, and operations, and therefore the overall bottom line, is many times greater.
Cross-functional teams
If you want to implement this kind of solution and get real value from it, you have to approach it as a transformation. “It’s not a simple solution like a copilot. Generative AI can improve your entire value chain. That’s where the real value is. That’s where the leaders are going to make the difference.”
For such a transformation to be successful, it takes creativity to connect these point solutions. That requires cross-functional teams. These include domain experts, people who are more on the strategic business side AND the specialists. “The reflex of traditional consulting firms in this regard is to first formulate a strategy, then build a business case, and from there realize a design. But when do you start delivering value? Before you know it, you are four months ahead of implementation. We turn that around. We do a short design sprint of two to four weeks. We look for the low-hanging fruit and build from there. During the build, we inevitably run into operational, strategic, and governance issues that we solve on the fly. That’s how we create value and embed it in the organization.
Where to start
The first step is to connect the dots between what is needed and what is possible. This can be done by creating a cross-functional team and taking a close look at the organization’s core tasks and processes. This allows you to map out where the semantic information about these tasks and processes is, and how (generative) AI can help turn this into actionable intelligence. Consider creating a customer journey and an employee journey. You can use these to identify where (generative) AI solutions can add real business value.
Do you want to take your AI project in your organization beyond a proof of concept? Get in touch.
