In Episode 23 of AI in NYC, hosts Rob May, Anna Kirk, and Ryan Eppley sit down with Gabi Steele, co-founder and CEO of Preql AI, to tackle one of the most overlooked problems in the AI gold rush: your data foundation is probably broken, and that's why your AI agents aren't working. Gabi brings a unique perspective — she went from building data visualizations at the Washington Post during Trump's first term to teaching at Columbia and Parsons, then dove deep into data engineering after watching her co-founder's experience at WeWork, where a 100-person data engineering team still couldn't wrangle the company's spiraling data infrastructure.

The episode kicks off with a deep dive into the bombshell WSJ report that OpenAI is missing key revenue and user targets ahead of its IPO. The panel debates whether this signals a real inflection point for the AI industry or just healthy competitive pressure from open-source models, DeepSeek, and a rapidly shifting market where agentic coding has overtaken conversational search as the killer app. Gabi weighs in with a grounded take from the data trenches — she talks to CFOs every day who tell her LLMs are too slow and they still want dashboards, revealing a massive gap between AI hype and enterprise reality.

The conversation evolves into a fascinating discussion about what interfaces should look like in an agent-first world. Rob shares his love for the physical Wall Street Journal, the group debates tactile vs. digital experiences, and Ryan introduces the concept of "Harness Experience" — designing not for users, but for agents. Gabi's core message resonates throughout: companies are rushing to deploy AI and automation but skipping the critical data layer that makes any of it actually work. If you're building with AI, buying AI tools, or leading a data team, this episode is essential listening.

Preql AI is building the infrastructure layer that reconciles and contextualizes enterprise data from across all your systems so AI agents can actually function. Currently focused on CFO buyers, Preql launched its agent-first product in July 2025. Learn more about Preql and connect with Gabi on LinkedIn.