#355 AI's Impact on Databases with Shireesh Thota, CVP of Databases at Microsoft

2026-04-13 · Source: DataFramed · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Cloud Computing & IT Infrastructure · Depth: Advanced, extended

Summary

Microsoft CVP of Databases, Shireesh Thota, discusses the evolution of data stacks, emphasizing the shift from fragmented modern data stacks to unified data platforms like Microsoft Fabric. Fabric integrates data integration, data science, data engineering, real-time analytics, and Power BI into a single SaaS-like environment with one security model and a unified data lake using open-source formats like Iceberg and Parquet. Thota highlights how AI agents are reshaping data interactions, enabling deeper reasoning with data through semantic models and ontologies. He also details Azure's database offerings, including Azure SQL, Azure Cosmos DB (a NoSQL database for internet-scale applications), Azure PostgreSQL, and Azure MySQL, explaining the trade-offs between SQL and NoSQL databases regarding consistency, scale, and availability. Microsoft's significant contributions to the PostgreSQL community and its open-sourcing of DocumentDB are also noted.

Key takeaway

For AI Engineers and Data Architects evaluating cloud data strategies, recognize that unified platforms like Microsoft Fabric streamline complex data stacks by integrating diverse services and leveraging open-source formats. While AI agents can generate queries, your understanding of SQL and data modeling remains critical for validating intent, optimizing performance, and ensuring data governance. Focus on building robust semantic models and ontologies to provide the necessary context for AI-driven reasoning, thereby enhancing application value and reducing operational overhead.

Key insights

Unified data platforms and AI agents are transforming data interaction, enabling deeper reasoning and simplified management.

Principles

• Data quality and context are paramount for effective AI reasoning.
• Unified platforms simplify integration, security, and governance.
• SQL and NoSQL databases serve distinct architectural trade-offs.

Method

Build a unified data layer with clean, curated data, then construct semantic models to define entities and relationships, and finally, create ontologies for business-specific policies and actions to enable deep AI reasoning.

In practice

• Utilize Microsoft Fabric for integrated data services.
• Consider Azure Cosmos DB for geo-distributed, high-scale NoSQL needs.
• Continue learning SQL for data modeling and intent validation.

Topics

• AI Agents
• Microsoft Fabric
• Azure Databases
• Data Stack Evolution
• Semantic Models & Ontologies

Best for: Data Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• Oracle converges the AI data stack to give enterprise agents a single version of truth — Oracle's converged database architecture aims to solve data fragmentation and consistency issues for enterprise AI agents.
• What Is a Data Agent? — Data agents transform data interaction from visual reports to natural language Q&A on governed datasets, streamlining insight delivery.
• AI needs a strong data fabric to deliver business value — Enterprise AI success hinges on data context and judgment, not just speed, necessitating a robust data fabric.
• Microsoft Build 2026: Pushing The Frontier With A More Opinionated AI Playbook — Microsoft's new AI strategy is a full-stack, opinionated approach, emphasizing distributed intelligence and robust governance.
• How World Bank Group uses databricks to eradicate poverty through shared knowledge — The World Bank Group unified disparate data on Databricks, using AI agents and RAG to democratize knowledge and accelerate global impact.
• From Context to Semantics: How Metadata Powers Agentic AI — Metadata platforms are evolving into semantic context layers crucial for precise AI and agentic system outcomes.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by DataFramed.