- Podcasts
- 09.09.2025
Special Episode: Digging into Generative AI
In this special episode...
We catch up with Eduardo Ordax, GenAI lead for AWS. We learn how little has really changed since 20 years ago Jeff Bezos used to say, 'Our most valuable asset right now is data.'
Building on the foundation that great data is the key to great AI, this podcast dives into the practical applications and challenges of generative AI for data professionals.
Learn how industry leaders are leveraging Retrieval Augmented Generation (RAG) to overcome model hallucinations and how the new wave of agentic AI is moving beyond simple chatbots to automate complex, multi-step workflows.
Discover key strategies for data leaders to build cross-functional teams, upskill their people, and move past the hype to deliver real business value with their AI initiatives.
Key takeaways
- Data as a foundational asset for AI success: Strong data foundations are crucial for successful AI implementation. AI is a consequence of data and cannot be done without it. AWS, for example, has historically leveraged data as its most valuable asset for customer recommendations and operational efficiency.
- Generative AI's democratization and practical applications: Generative AI has significantly democratized access to AI by providing widely understood and valuable use cases for both business and personal use. It has moved beyond narrow applications to encompass diverse functions like enhancing customer experience (chatbots), boosting productivity (code writing, RAG) and automating internal workflows.
- Importance of high-quality, timely data for RAG: For Retrieval Augmented Generation (RAG) solutions, the quality and timeliness of data are paramount. Eduardo highlights that issues with data, such as being outdated or unclean, can lead to incorrect answers from models, even if the model itself is not the problem. Effective RAG requires clean, updated, unbiased, and clearly structured information.
- Complexity and testing in RAG implementations: Implementing RAG at scale is complex and requires extensive testing with a significant amount of data, not just small samples. Strategies for structuring information (e.g. chunking techniques), experimenting with different retrieval methods (semantic vs. keyword search) and defining a robust vector database strategy are critical for optimal results.
- People, processes and cross-functional collaboration are key: The success of AI initiatives, including agents, is less about technology and more about having the right team, processes and strong collaboration between IT and business. Data leaders must invest in upskilling their teams and fostering cross-functional teams to break barriers and work together effectively on use case implementation and validation.
Also available on major podcast platforms and Matillion's YouTube.
We catch up with Eduardo Ordax, GenAI lead for AWS. We learn how little has really changed since 20 years ago Jeff Bezos used to say, 'Our most valuable asset right now is data.'
Building on the foundation that great data is the key to great AI, this podcast dives into the practical applications and challenges of generative AI for data professionals.
Learn how industry leaders are leveraging Retrieval Augmented Generation (RAG) to overcome model hallucinations and how the new wave of agentic AI is moving beyond simple chatbots to automate complex, multi-step workflows.
Discover key strategies for data leaders to build cross-functional teams, upskill their people, and move past the hype to deliver real business value with their AI initiatives.
Key takeaways
- Data as a foundational asset for AI success: Strong data foundations are crucial for successful AI implementation. AI is a consequence of data and cannot be done without it. AWS, for example, has historically leveraged data as its most valuable asset for customer recommendations and operational efficiency.
- Generative AI's democratization and practical applications: Generative AI has significantly democratized access to AI by providing widely understood and valuable use cases for both business and personal use. It has moved beyond narrow applications to encompass diverse functions like enhancing customer experience (chatbots), boosting productivity (code writing, RAG) and automating internal workflows.
- Importance of high-quality, timely data for RAG: For Retrieval Augmented Generation (RAG) solutions, the quality and timeliness of data are paramount. Eduardo highlights that issues with data, such as being outdated or unclean, can lead to incorrect answers from models, even if the model itself is not the problem. Effective RAG requires clean, updated, unbiased, and clearly structured information.
- Complexity and testing in RAG implementations: Implementing RAG at scale is complex and requires extensive testing with a significant amount of data, not just small samples. Strategies for structuring information (e.g. chunking techniques), experimenting with different retrieval methods (semantic vs. keyword search) and defining a robust vector database strategy are critical for optimal results.
- People, processes and cross-functional collaboration are key: The success of AI initiatives, including agents, is less about technology and more about having the right team, processes and strong collaboration between IT and business. Data leaders must invest in upskilling their teams and fostering cross-functional teams to break barriers and work together effectively on use case implementation and validation.
Also available on major podcast platforms and Matillion's YouTube.
Share: