How Xeneta is using AI to transform how freight is bought and sold

No matter where you look, it’s hard to escape AI today. It’s being applied in virtually every function and process in modern businesses — and procurement, supply chain, and logistics are no exceptions. 

As a megatrend, it’s about as hot as hot topics can get. But at Xeneta, it’s something we’ve been using for years as part of our core platform architecture. For example, we use a combination of machine learning, predictive analytics, analyst expertise and community insights to help us: 

• Classify rate sheets 

• Automate data processing 

• Detect outliers and spot anomalies in our data 

• Predict future market movements in our Market Rate Outlook product 

It took Xeneta a lot of time, testing, and experimentation to find the perfect use cases for AI in its products and processes. In this article, I’ll give you an inside look at how we settled on those decisions, and the approach that’s going to help us continue applying AI in new ways to transform how freight is bought and sold. 

Xeneta's AI journey is driven by its customers 

While it’s undeniably powerful, integrating AI into existing products isn’t valuable by default. In fact, adding “AI for AI’s sake” can have a serious detrimental impact on customer experiences with a product. All it takes is one hallucination, and suddenly a customer can completely lose trust in your data. 

So, when Xeneta first starts exploring potential use cases for AI in its products, we always start by asking customer-centric questions like: 

• What are our customers’ most persistent and significant challenges? 

• How could AI help Xeneta simplify how customers engage with our products? 

• What are the key points of friction or inefficiency in our customers’ daily workflows? 

It’s our view that AI is not inherently valuable for our customers. Its value is unlocked through customer-centric applications and use cases. It’s not a feature in its own right, but a technology that enables us to solve our customers’ problems in new ways. 

Photos taken from Summit 2024, Amsterdam

With AI, data quality is paramount 

The biggest pitfall with AI — particularly generative AI (GenAI) — is that what you get out of it is dependent on the quality of the data you put in. All models are still prone to errors and hallucinations, which present a serious issue for applications around the buying and selling of freight. 

Xeneta has spent many years building a strong reputation as the leading provider of timely, reliable, and accurate freight buying and selling data. Our customers trust the data and insights we provide, and use them to make high-value decisions every day. Providing incorrect or incomplete data can cost our customers millions of dollars. Quality is therefore something that we are simply not willing to compromise on. 

As our data is also sourced from our customers, we also fully anonymize it and ensure that anything which could potentially identify or be linked back to a specific customer is never fed into any kind of AI model. 

We’re excited about the potential of AI to unlock new forms of insight and personalization — including the ability to surface predictive or estimated data tailored to individual users. But our bar remains high. Any AI-generated output must meet the same standard of trust and reliability that our customers expect from Xeneta. We’ll never compromise on data integrity — and every AI application we introduce will be rooted in real-world freight context, not generic models, and will be scrutinized by our human experts. 

But that doesn’t mean AI can’t help us evolve our products and processes. Far from it, in fact. 

Our commitment to continuous experimentation 

The only way to find the most-impactful areas where AI could deliver the greatest customer value is to continuously experiment with new models and use cases. This is critical not only for bringing new features to our products, but also for discovering which popular capabilities might not be right for us and our customers. 

For example, last year, we experimented with an AI chatbot answering rate-related questions. After testing the idea with over 100 customers, we decided not to add to our product just yet. Most of the questions that the chatbot was asked were better answered with a different visualization. While this experiment didn’t end in the addition of a new feature, it did help us learn about the types of questions our customers ask — and by extension, what they really want from Xeneta. 

This has previously led to the creation of new insights such as Peer Comparison by Volume, where shippers can compare themselves with customers shipping similar volumes on a specific port-port route. 

Now, we’re using that insight to drive further experiments. For example, we’re building personalized insights to inform shippers and logistics service providers about opportunities to optimize their supply chain. If we spot that a port-port rate has become uncompetitive compared to the market for instance, we can prompt the customer with a recommendation to renegotiate. 

AI also has an important role in the product development process. For example, we are using large language models to rapidly create prototypes of new solutions and features, and get those prototypes into customers’ hands to test their viability. A process that used to take weeks or months can now be completed in just hours or days, which helps us cover a huge amount of experimental ground and learn a lot in a really short space of time. 

Similarly, we also use AI internally to help transform processes and keep productivity as high as possible. For example, we use GenAI tools to help accelerate workflows for our engineering teams. This ensures that our people can spend more time focusing on customer-centric innovation rather than simple routine tasks, and contribute as much as possible to the continuous improvement of our products and customer experiences. 

What’s next: Evolving from data to insight delivery 

Historically, our customers have always appreciated our approach to collecting and aggregating freight data. We’ve provided the data our customers need to make informed, timely, and optimized freight buying and selling decisions, but there’s sometimes still a level of effort needed on their side to translate that data into contextualized insight. 

Looking ahead, that’s one of our biggest focus areas. We want to remove that complexity and provide our customers — and each customer’s individual users — with clear, actionable insights.  

AI will be a crucial enabling technology as we progress on our mission to transform how freight is bought and sold. Through continuous innovation, experimentation, and customer validation, we’ll ensure we always apply it in ways that amplify the customer impact we deliver through our data and never risk undermining it. 

I can’t wait to share more AI-enabled features with you in the future.