AI and Knowledge: Pablo Castro on Microsoft’s AI Knowledge Systems

Highlights

04:51

Model parameters should not be the primary storage of knowledge; retrieval systems are key.

05:03

RAG is not a temporary patch but a long-term architectural choice that needs deep integration with agents.

08:26

Foundry IQ can automatically understand documents, tables, and images, solving multimodal knowledge understanding issues.

⭐⭐⭐⭐ 4.2

This episode features Pablo Castro, a Distinguished Engineer and CVP at Microsoft leading the AI Knowledge team within CoreAI.

The conversation explores how AI and knowledge systems intersect to build better applications and agents, covering the key systems his team develops: Foundry IQ, Azure AI Search, and Azure Content Understanding.

Castro discusses the challenges of state-of-the-art information understanding and retrieval, and how these capabilities enable more effective AI-powered experiences.

The episode is aimed at engineers and product managers interested in practical knowledge retrieval architectures for AI agents.

This episode features Pablo Castro, a Distinguished Engineer and CVP at Microsoft leading the AI Knowledge team within CoreAI.

The conversation explores how AI and knowledge systems intersect to build better applications and agents, covering the key systems his team develops: Foundry IQ, Azure AI Search, and Azure Content Understanding.

Castro discusses the challenges of state-of-the-art information understanding and retrieval, and how these capabilities enable more effective AI-powered experiences.

The episode is aimed at engineers and product managers interested in practical knowledge retrieval architectures for AI agents.

On AI and Knowledge — Pablo Castro, Distinguished Engineer & CVP for AI Knowledge, Microsoft

View Original