958: Without Trusted Context, Agents are Stupid (featuring Salesforce's Rahul Auradkar) - Super Data Science: ML & AI Podcast with Jon Krohn Recap

Podcast: Super Data Science: ML & AI Podcast with Jon Krohn

Published: 2026-01-16

Duration: 24 min

Guests: Rahul Auradkar

Summary

AI models often fail to perform effectively because they lack the necessary context. Rahul Auradkar explains Salesforce's approach to providing trusted context through a unified data engine, enhanced by the acquisition of Informatica.

What Happened

Rahul Auradkar, EVP and GM at Salesforce, delves into the importance of providing AI models with trusted context to enhance their effectiveness. He highlights that while AI models are intelligent, they often lack the context needed to make informed decisions, which can lead to suboptimal outcomes.

Salesforce's unified data engine is a solution designed to provide this context by integrating various components such as Data360 and Tableau. Data360, formerly known as CDPs, focuses on unlocking trapped data and harmonizing it to provide actionable insights.

Tableau, acquired by Salesforce in 2019, is lauded for its intuitive analytics capabilities and is now integrated with the Salesforce platform to offer agentic analytics. This integration involves a semantic data model that ensures a consistent understanding of enterprise data.

MuleSoft, another component of the unified data engine, facilitates app-to-app integration and API management, providing governance and orchestration for agent actions within enterprises.

The episode also highlights Salesforce's acquisition of Informatica, which brings data management tools like data quality and integration capabilities, enhancing the unified data engine's ability to provide trusted context.

Auradkar emphasizes that enterprises are often data-rich but context-poor, leading to ineffective AI applications. By providing a unified data context, AI agents can deliver more personalized and meaningful experiences.

The discussion concludes with insights on how enterprises can succeed with agentic AI by 2026, using Salesforce's internal implementation as an example of how agentic transformations can be achieved.

Key Insights

• AI models require trusted context to function effectively, as they often lack the information needed to make informed decisions, leading to suboptimal outcomes.
• Salesforce's unified data engine integrates components like Data360 and Tableau to provide a consistent understanding of enterprise data, enhancing AI capabilities.
• MuleSoft facilitates app-to-app integration and API management within Salesforce's platform, ensuring governance and orchestration for enterprise agent actions.
• Salesforce's acquisition of Informatica enhances data management with tools for data quality and integration, further supporting the provision of trusted context for AI agents.