Data is the Differentiator for Successful Gen AI Adoption

A joint webinar featuring experts from Amazon Web Services (AWS) and Windward had a recurring theme: data is the foundation and differentiator when it comes to successful Gen AI adoption. Our first blog post in this series highlighted other adoption challenges as well. 

Vadim Tereshchuk, Solutions Architect at AWS, stressed that Gen AI applications are just the tip of the iceberg, the part that customers see – but these applications are built on a lot of data behind the scenes. 

“When you start speaking about Gen AI, it’s all about the data,” added Eric Topp, Head of Worldwide Solutions & Strategy, Transportation & Logistics, AWS.  

“There’s this big rush to do something with Gen AI from leadership, but if you don’t have the right foundation and cleanse your data, harmonize it, and hydrate your data lake properly, you can take a really good Gen AI application and turn it into something that’s worthless, because your data is not good. You can ask a question, and you’ll get an answer. It just might not be the right answer because the data is incorrect.” 

A top-down approach is now needed. In the shipping world, many organizations have totally relied on buying data from third-party data sellers. The Gen AI conversation flips this story. Instead of merely asking, “Which data can you give me?” or “Can you tell me where this ship is going and what it’s doing?”, the focus has shifted to, “Which problem are we really trying to solve with the sea of data that exists?” 

Organizations are taking a fresh look at data lakes, applications, and needs.

Curated data is the cure – this approach combines domain knowledge and advanced technology. Instead of relying solely on raw data, Windward uses labeled datasets, customer data, and quality assurance processes to train our vertical AI models. Vertical AI generates domain-specific insights and answers, which is critical in the complex, interdependent shipping industry. This overall methodology ensures higher accuracy and reliability, meeting the detailed demands of the maritime industry.

Structured vs. Unstructured Data

Trust in AI output has been a significant topic of discussion, underscoring the importance of reliable and transparent AI systems, and experienced vendors. Clean, harmonized, and well-hydrated data lakes are essential. Without a solid data foundation, and an ability to deliver the right insights at the right times, even the most advanced AI applications can deliver inaccurate or misleading results, undermining their potential benefits. 

Effective technology must be able to handle and process structured and unstructured data. 

Structured data – such as vessel schedules, AIS data, and freight rates in the ocean logistics ecosystem – is organized in predefined formats and typically stored in databases. This organization facilitates systematic access, querying, and analysis, and supports the development of AI models.

Unstructured data – including text documents and contracts, such as digital bills of lading (BoL) and port contracts between freight companies and carriers, and between freight companies and their customers – requires more sophisticated handling. 

This often involves using vector databases and embedding models to filter and extract relevant information. Managing unstructured data is a daunting task because of the convoluted processes involved. Despite the enormous potential for business growth, many companies are deterred by the time-consuming nature of these processes.

Integrating unstructured data into vector databases using embedding models helps filter and extract valuable information from text documents. Unstructured data can improve Gen AI’s performance, as it provides more context and data for training the models.

Our NEW eBook, The Natural Evolution of Gen AI for the Maritime Ecosystem, explains how to move beyond the Gen AI buzz, exactly which pain points Windward’s Gen AI agent heals, and much more!