Your LLM issues are really data issues

About this episode

Ryan discusses with Harsha Chintalapani the challenges AI and LLMs face with real-time, structured production data and how to prepare data for AI.

Ryan welcomes Harsha Chintalapani, co-founder and CTO at Collate and co-creator of Open Metadata, to the show to discuss why AI and LLMs struggle with real-time, structured production data. They explore how schema changes, inconsistent definitions (like “customer”), and weak governance can break both your analytics and MLs, and what companies can do to get their data AI-ready, from metadata management to observability.  Episode Notes:   Collate is a semantic intelligence platform built on a semantic metadata graph for discovery, governance, and AI observability across your data ecosystem. Connect with Harsha on LinkedIn .  Congrats to user buttonsrtoys , who won a Famous Question badge for their question Possible to edit PDF without embedded font installed? . TRANSCRIPT See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info .

People in this episode

Host: Ryan

Guest: Harsha Chintalapani

Topics covered

• AI
• LLMs
• data governance
• metadata management
• analytics
• machine learning

Keywords

• AI
• LLMs
• data issues
• metadata
• analytics
• machine learning
• governance

Mentioned in this episode

Organizations: Collate, Open Metadata

More episodes of The Stack Overflow Podcast

• Developers are emotionally attached to their tools · June 12, 2026 · 34 min
• When the cost of code approaches zero, what does engineering leadership look like? · June 11, 2026 · 34 min
• Creating checkpoints by gaslighting a Postgres database · June 9, 2026 · 35 min
• Making the OWASP top ten in the vibe code era · June 5, 2026 · 34 min
• What it takes to be a player in the international AI game · June 2, 2026 · 26 min
• The find out stage of AI is just supply chain and password protection · May 29, 2026 · 31 min