For many companies, Artificial Intelligence (AI) seems like a magic word these days. But how do you make it applicable in daily practice? That's what Progent's programmers focus on every day. "How cool is it when AI actively alerts you that a customer who usually orders strawberries every Tuesday hasn't done so this week!"
Progent specializes in turning data into actionable insights for the fresh produce sector. The company counts major fresh produce trading companies among its clients. "We combine data from ERP systems, financial software, production machines, and truck on-board computers into a Data Warehouse. Then we create dashboards that provide an at-a-glance view of company performance," explains William Verhoeven.
With AI, the company takes it a step further—but there's an important prerequisite. "A company needs to have its data management in order. If that foundation is solid, we can use AI to connect analyses, forecasts, and calculations," explains Mark Verloop. "This allows businesses to identify missed sales, for example quickly. I am convinced this can generate substantial revenue for many companies."
William Verhoeven And Mark Verloop of Progent with their first version of the AGF chatbot
Reading PDFs with price lists
"A next step for AI, where it is becoming increasingly intelligent, is processing unstructured data, such as emails with pricing information. Currently, many companies start their day benchmarking prices using lists often sent as PDFs. These prices are still frequently retyped manually. However, AI can now be deployed to read PDFs or Excel files and compare prices effectively."
"Another step is answering customer queries from the system, such as when a truck will arrive," Verloop continues. Asked if AI is already capable of this, the programmer responds, "It's still AI, and it's not perfect in every area, but these modules learn incredibly fast. The key is for companies to set clear goals so they can expect the model to deliver the desired answers. We are now developing a chatbot specifically for the fresh produce sector. That's truly unique in the market."
When asked what AI cannot yet do, Mark replies, "AI is in full development. Applications that weren't feasible six months ago are now becoming possible. However, it's not as simple as flipping a switch to get a predictive pricing model you can rely on blindly. For instance, you can hypothesize that poor weather in Costa Rica will result in a bad season and affect availability and prices, but it's still just a forecast—not an absolute truth. No one could have foreseen the extent of storm DANA in Spain, for example."
Recognizing trends
"Currently, Progent's AI chatbot can already answer many questions: Which customer had the highest sales last month? Who was the biggest pomegranate buyer? And who filed the most claims? Questions that would normally require consulting an accountant can now be answered in real-time, with an emphasis on spotting trends," says William. "But it's not plug-and-play; a structured data model is necessary for proper configuration."
For Progent's customers, the AI module is almost ready. "We're putting on the finishing touches and will be ready for release next year. That's why we're already preparing the market for the possibilities AI offers. The customers we serve already have a Data Warehouse, so they'll be able to start using it right away!"
For more information:
William Verhoeven
Progent
Tel: +31 (0)85 82 24 12 4
[email protected]
www.progent.nl