What makes a good AI benchmark? Greg Kamradt joins Demetrios to break it down—from human-easy, AI-hard puzzles to wild new games that test how fast models can truly learn. They talk hidden datasets, compute tradeoffs, and why benchmarks might be our best bet for tracking progress toward AGI. It’s nerdy, strategic, and surprisingly philosophical.

I’m sure the MLOps community is probably aware – it's tough to make AI work in enterprises for many reasons, from data silos, data privacy and security concerns, to going from POCs to production applications. But one of the biggest challenges facing businesses today, that I particularly care about, is how to unlock the true potential of AI by leveraging a company’s operational business data. At Snow Leopard, we aim to bridge the gap between AI systems and critical business data that is locked away in databases, data warehouses, and other API-based systems, so enterprises can use live business data from any data source – whether it's database, warehouse, or APIs – in real time and on demand, natively. In this interview, I'd like to cover Snow Leopard’s intelligent data retrieval approach that can leverage business data directly and on-demand to make AI work.

The creator of FastAPI is back with a new chapter—FastAPI Cloud. From building one of the most loved dev tools to launching a company, Sebastián Ramírez shares how open source, developer experience, and a dash of humor are shaping the future of APIs.

Willem Pienaar and Shreya Shankar discuss the challenge of evaluating agents in production where "ground truth" is ambiguous and subjective user feedback isn't enough to improve performance. The discussion breaks down the three "gulfs" of human-AI interaction—Specification, Generalization, and Comprehension—and their impact on agent success. Willem and Shreya cover the necessity of moving the human "out of the loop" for feedback, creating faster learning cycles through implicit signals rather than direct, manual review. The conversation details practical evaluation techniques, including analyzing task failures with heat maps and the trade-offs of using simulated environments for testing. Willem and Shreya address the reality of a "performance ceiling" for AI and the importance of categorizing problems your agent can, can learn to, or will likely never be able to solve.

Tecton Founder and CEO Mike Del Balso talks about what ML/AI use cases are core components generating Millions in revenue. Demetrios and Mike go through the maturity curve that predictive Machine Learning use cases have gone through over the past 5 years, and why a feature store is a primary component of an ML stack.

Memory is one of the thorniest challenges in deploying LLMs. Mem0 introduces a scalable long-term memory architecture that dynamically extracts, consolidates, and retrieves key information from conversations. By using graph-based structures to model relationships between conversational elements, Mem0 enables AI agents that are more accurate, coherent, and production-ready.

Raza Habib, the CEO of LLM Eval platform Humanloop, talks to us about how to make your AI products more accurate and reliable by shortening the feedback loop of your evals. Quickly iterating on prompts and testing what works, along with some of his favorite Dario from Anthropic AI Quotes.

An analysis of the May 2025 incident where xAI's Grok chatbot began inappropriately referencing 'white genocide' in South Africa. This post-mortem delves into the probable cause—a flawed post-processing prompt—framing it as a critical MLOps failure. It underscores the necessity of treating prompts as key artifacts, implementing progressive deployment strategies, and using appropriate metrics for AI safety and reliability.