Databricks and Infosys on Rebuilding Data Infrastructure for Enterprise AI
In a podcast episode of MIT Technology Review's Business Lab, Bavesh Patel (SVP at Databricks) and Rajan Padmanabhan (Unit Technology Officer at Infosys) discuss the critical need for enterprises to rebuild their data infrastructure to successfully adopt AI. They argue that fragmented, siloed data is the main obstacle to meaningful AI deployment, and that organizations must consolidate data into open formats, implement governance, and tie AI initiatives to business metrics. Patel highlights Databricks' Lakehouse and new Lakebase OLTP database for AI agents, while Padmanabhan emphasizes a six-part framework for value measurement and governance. They predict a shift from copilots to autonomous agents, enabling new business models and efficiencies.
Key facts
- Bavesh Patel is SVP at Databricks.
- Rajan Padmanabhan is Unit Technology Officer at Infosys.
- The podcast is produced in partnership with Infosys Topaz.
- Patel states that AI quality depends on enterprise data quality.
- Padmanabhan says enterprise AI requires precision above 92%.
- Databricks launched Lakebase, an OLTP database for AI agents.
- Unity Catalog is Databricks' governance product.
- Patel notes that 95% of AI projects fail to generate business value.
- Padmanabhan describes a six-part framework including value management and governance.
- The podcast host is Megan Tatum from MIT Technology Review.
Entities
Institutions
- Databricks
- Infosys
- Infosys Topaz
- MIT Technology Review
- Business Lab
Locations
- Brighton
- England