Interviewer: Welcome to The AI Summit London. We're delighted to be joined by Sourav Das, Senior Platform Product Manager at AP Moller Maersk

So thinking about whether in the US or South Africa, what are you building and what problem are you solving for procurement managers across your global footprint?

Sourav: I am building a unified and standard platform that allows procurement managers across all 130 countries to see their category strategy taking shape. It will be dynamic and respond to live market conditions rather than sitting static in a PowerPoint on someone’s desktop.

Interviewer: How does the category strategy assistant change a category manager’s workflow, and can you share a practical example?

Sourav: The need came from how strategy work was being done. Most procurement managers created a strategy once a year and started from scratch each time. They had to move across fragmented systems, which took significant time and effort. Because of that, some countries did not adopt a category strategy at all. Our aim is to offer a platform that condenses six to eight weeks of research into a few hours. By selecting options in the portal, managers can view country specific recommendations before they enter the sourcing stage. Time savings were the key driver.

Interviewer: As AI moves from information gathering to decision making, what shifts are you seeing in knowledge work, and why do leaders sometimes underestimate the change?

Sourav: We have veterans at Maersk with 10 to 15 years of experience who know their markets very well. Without tools powered by AI, most of their time goes into gathering information and research. With AI and large language models, that work becomes much easier. Today, about 60 to 70 percent of the time goes into gathering information and 30 percent into decision making. We want to reverse that so teams spend minimal time on gathering and much more on interpreting insights and making decisions. That is where they add the most value.

I would not say leaders underestimate the change. Many legacy companies move carefully because change management matters. Leaders do not want to push too hard on frontline teams where there can be a fear that if AI does certain tasks, people will lose their jobs. Addressing that fear and pacing the change are essential.

Interviewer: Many conversations focus on the challenges of adopting and scaling AI across organisations. What challenges are you facing, and how are you addressing them?

Sourav: Three things stand out.

First, adoption. You need a trust loop with users. If you hand them a portal and say do what AI recommends, they will not do it. From day one we identified change agents and champions among key business users and involved them in building the product. They understand how the model works and they help shape it. We also add transparency features such as confidence scores and explainability so users see why a recommendation was made.

Second, data availability. No AI works without data. We prioritised categories where we had strong data coverage. Otherwise, the outcomes would not be reliable.

Third, success metrics. For an AI product, model accuracy is useful but not the main goal. What matters is business outcome. Are users more productive. Are we achieving cost savings and dollar savings. Those are the metrics that count.

Interviewer: Based on what you have learned, what practical advice would you give leaders to ensure successful generative AI implementation in their organisations?

Sourav: First, adoption. Identify champions among your core business users, involve them early, and make them part of the end to end journey. Second, data. Focus on domains where you already have good, structured data rather than trying to fix data as an afterthought. Third, metrics. Do not anchor only on technical metrics. Prioritise business outcomes and the benefits users gain, such as productivity and cost savings.

Interviewer: Looking ahead to the next 12 months, what trends are you watching or planning to implement in the generative AI space?

Sourav: Things are moving fast. A year ago, people were excited about prompting and using large language models for questions, and AI agents sounded like a future idea. Now agents are already active in production. It is not theory. Agents are taking actions, not only making recommendations. Everyday users might rely on chat tools for queries, but industries, including legacy sectors, are now factoring agents into their operating models. The next phase is about agents doing work.

The second trend is adoption capacity. The constraint is less about technology readiness and more about user readiness, inside and outside the company. The pace is intense. With each new model or cloud release, tech moves forward, but change management is more important than ever. Users must keep up, and that gap needs to be addressed quickly.

Third, generative AI is moving deeper into supply chain and procurement. What started on the consumer side is now shaping strategic sourcing, supply chain technology, and autonomous agent to agent negotiations within the enterprise. It is no longer limited to consumer experiences. It is changing business to business operations.

It is no longer limited to only the consumer; it has moved into the B2B side of things as well.

Interviewer: So they are now getting a more 360 degree model. Fascinating. It’s your first time at The AI Summit London. What has your experience been so far?

Sourav: It has been a fascinating and genuinely practical experience. I would love to come back. I have attended many talks and explored the expo, and I see a lot of practitioners who are building, learning, and openly sharing what works and what does not. It is no longer theory in motion. You can see products in practice and real applications, which is far more convincing than a theoretical lecture. The same goes for my panel on the applied AI track, where we focused on how industry is deploying AI today. Learning from companies and peers about how they are applying it is far more enriching than a purely theoretical discourse.

Closing

From compressing weeks of research into hours to shifting human effort from information gathering to decision making, Sourav’s focus is on practical impact, trust, and measurable outcomes. 

As AI agents move from concept to execution and generative AI scales across supply chains and sourcing, success will belong to teams that engage users early, invest in the right data, and measure what matters.