AI Enterprise - Databricks & Glean | BG2 Guest Interview

What It Covers

Databricks CEO Ali Ghodsi and Glean CEO Arvind Jain discuss enterprise AI adoption realities, explaining why 95% of AI projects fail, where economic value accrues, and how their companies automate workflows across finance, healthcare, and retail sectors.

Key Questions Answered

• •LLM Commoditization: Large language models function as interchangeable commodities like gas stations. Companies switch between OpenAI, Anthropic, and others weekly based on price and marginal performance differences. Competitive advantage comes from proprietary company data and business processes, not model selection.
• •AGI Already Exists: Current AI capabilities meet the artificial general intelligence definition used by researchers for thirty years. The industry moved goalposts after achieving original AGI benchmarks. Enterprises need to focus on making existing AI useful rather than waiting for superintelligence breakthroughs.
• •Enterprise Success Pattern: Royal Bank of Canada reduced equity research report generation from two hours to fifteen minutes using agents that analyze earnings calls, competitor data, and market news. Seven Eleven automated marketing segmentation and content creation. Success requires engineering effort, evaluations, and strong teams, not quick implementations.
• •Speech Interface Revolution: Keyboards will disappear as speech becomes the primary interaction method. Data entry shifts from manual form filling to conversational input captured through tools like Zoom. Meeting recordings automatically update CRM systems and knowledge bases, eliminating structured data entry workflows entirely.