Description

In this episode of Chain Reaction, Michael Ostroumov welcomes Bram Vanschoenwinkel, Chief Product Officer at Customaite and a longtime AI practitioner, to explore the practical impact of artificial intelligence on supply chain and customs operations. Bram provides a compelling look at how AI has evolved over decades, far beyond the recent hype surrounding generative models, and how these advancements are now delivering tangible value in logistics.

Bram delves into the role of agentic AI, which moves beyond reactive tasks to proactive decision-making, transforming traditional customs declaration workflows. He shares how Customaite leverages AI to automate and accelerate customs processes, reducing hours of work to minutes without eliminating human oversight. This human-assisted AI approach safeguards against errors and bias, fosters trust, and enhances productivity - key to empowering supply chain professionals in a rapidly digitizing world.

Guest Bio

Bram Vanschoenwinkel is the Chief Product Officer at Customaite, a company transforming customs declaration through AI. With a PhD in machine learning and decades of experience in artificial intelligence, Bram specializes in natural language processing and supply chain automation. His work focuses on leveraging agentic AI to enhance productivity while maintaining human expertise at the core of decision-making. Under his leadership, Customaite has pioneered innovative approaches that significantly reduce customs processing time while preserving data integrity and operational accuracy.

Takeaways

• AI in supply chains extends far beyond generative models - most real-world systems still use classical machine learning.
• Agentic AI introduces proactive, multi-step task management, revolutionizing logistics operations.
• Customaite uses AI to analyze diverse document formats and generate customs declarations.
• Human-assisted AI ensures decision transparency and learning from human corrections.
• Clear UI with traceability markers boosts user confidence in AI-derived decisions.
• Data validation in Customaite is structured through historical data, master data, and user feedback.
• A key use case is reducing document preparation from 8 hours to just 20 minutes.
• Customaite primarily serves logistics providers and customs brokers across 12 European countries.
• Maintaining data privacy and using anonymized data helps train stronger, safer models.
• Conservative supply chain players are adopting cloud and AI solutions with clearer ROI evidence.
• Measuring time saved per tariff line helps quantify AI's business value.
• AI won't replace jobs - but those who learn to use AI will replace those who don’t.

Chapters

00:00 Introduction to Bram and AI in supply chains

01:18 Overview of AI history and current use cases

03:09 Generative vs. agentic AI in logistics

05:22 Human-machine collaboration and the "Skynet" myth

07:28 Customaite's philosophy: human-assisted AI

09:50 The need for human expertise in edge cases

12:21 Customaite’s specific use case in customs declarations

13:54 How Customaite processes documents and adds context

16:25 Who Customaite serves and how they differ from other platforms

18:13 The “digital twin” of a customs declarant's workspace

19:51 Real-world time savings: from 8 hours to 20 minutes

22:17 Managing trust and privacy in AI systems

24:30 Anonymized data use to train robust models

26:50 Selling AI to conservative logistics players

28:35 Measuring ROI with a pay-per-use model

⁠⁠Follow Bram Vanschoenwinkel on LinkedIn⁠⁠⁠

⁠⁠⁠Follow Michael Ostroumov on LinkedIn