From AI experiment to scalable solution: how do you take concrete steps?
More and more organizations are ready for the next phase with AI. They experimented with tools such as Copilot or ChatGPT, set up the first minimum viable products (MVPs) and ran proofs of concept (POCs).
Many AI experiments lead to enthusiasm. But as soon as an experiment goes well and you want to take it bigger, new challenges arise. Think of governance, compliance, security, data quality and adoption. So the question is: how do you scale AI in a responsible and effective way?
We sat down with Megan Bloemsma, AI lead at Wortell. In this blog post you can read her vision and advice!
First step towards long-term thinking: include company data
"I notice that many organizations have reached a tipping point," says Megan. "They want to see AI as a long-term topic and make it practical. But it is difficult to determine which path is best for them to take."
Megan's advice? Start by taking company data with you. She illustrates what she means by this with an example: "Suppose you ask Copilot or ChatGPT what trends are coming and how you can respond to them. If you specify the country or region, you get quite a nice answer. But what if you enter: 'What trends can we expect based on our CRM data and notes of customer conversations?' Then the answer becomes much more specific, because the AI tool can filter which opportunities exist for you and your customers within the broader range of trends."
From preparations to practical insights
Including company data is a good first step to setting up a scalable AI solution. But where do you start? "It's important to add your data to AI securely and compliantly ," Megan explains. "To do that, you first want to gain insight into the data you have. Then you clean everything up: you classify or recategorize data. This lays a solid foundation from which you can further optimize."
Do you need to make more preparations to roll out an AI solution more widely in the organization? "That's possible," says Megan, "but I also believe that you just have to start somewhere. You can make a thousand preparations and not be much wiser. Ultimately, practical insights take you further. When you have laid a safe foundation, it is often wise to test how everything works in practice. You do this step by step. For example, after adding company data, take a look at what output your AI tool now provides on a somewhat larger scale. Sometimes you see that only one source, such as sales data, offers real enrichment. Then you give them priority. You may find out that the state of data is worse than you thought. In that case, you can put improving data quality at the top of the agenda."
Pushing for a positive domino effect
In the rapidly developing landscape, you should no longer see the rollout of an AI solution as a one-off project. Megan: "It's an ongoing process. With every step you take, you get new insights. You use these to further optimize. If you persevere, you will see a positive domino effect emerge. My advice is therefore: start somewhere, make sure you use a safe approach and know that this process has no end date!"
Want to get off to a good start? Test your AI maturity with the AI Scan to get a clear picture of where you stand and where your improvement potential lies.