The Data Exchange with Ben Lorica

Yoni Leitersdorf, the CEO and co-founder of Solid, dives into the world of AI and data access. He explains why simply using LLMs for text-to-SQL can lead to errors without adequate business context. Yoni highlights the importance of a semantic layer, which acts as a Rosetta Stone for AI, providing essential data quality and query context. He discusses how organizations can start building this layer, the role of human oversight in AI accuracy, and the challenges in creating cross-system semantic layers. Plus, he shares his vision for the future of AI in enterprise.

Evangelos Simoudis, a venture investor and AI strategy expert, dives into the disconnect between the hype surrounding AI agents and their actual enterprise application. He highlights critical issues like the reliance on shadow agents, the necessary foundations for effective AI, and the importance of clean data. Topics include the challenges faced in pilot programs and the need for reliable AI solutions. They also discuss the potential negative impact of new H-1B visa policies on tech startups, advocating for a balanced approach to immigration reform.

Jakub Zavrel, CEO of Zeta Alpha and expert in AI research agents, dives into the evolution from traditional enterprise search to advanced deep research systems. He contrasts user expectations of fast web searches with the thorough, report-style answers of deep research. Zavrel discusses the importance of customization and quality optimization, and his innovative JEPA method for prompt evolution that outperforms traditional reinforcement learning. He also emphasizes the need for secure enterprise search as a foundation for powerful AI solutions.

Andrew Rabinovich, CTO and Head of AI at Upwork, dives into the synergy between digital work marketplaces and AI agents. He reveals how Upwork's unique Reinforcement Learning from Experience (RLEF) methods optimize agent training. The discussion highlights the importance of user evaluation in shaping AI performance and explores innovative multimodal feedback systems. Rabinovich also envisions a future where human workers and AI coalesce seamlessly, providing expedited services and new opportunities for freelancers.

Jay Alammar, Director and Engineering Fellow at Cohere and co-author of "Hands-on Large Language Models," delves into enterprise AI. He discusses the challenges in understanding large language models and the adoption of GraphRag, emphasizing the gap between vendor enthusiasm and real-world application. Alammar highlights the balance between self-directed and collaborative learning in AI, and the critical role of evaluation processes for effective AI development. He also explores the potential of smaller AI models, showcasing their efficiency in addressing specific tasks.

Evangelos Simoudis, a venture capitalist at Synapse Partners and a corporate innovation blogger, dives deep into AI's complex landscape. He reveals the challenges of AI regulation in the U.S. compared to China's engineering-centric approach. The discussion highlights how China's governance fosters rapid technological advancement. They also explore issues around intellectual property in training data and the need for transparent regulations. Simoudis emphasizes the importance of balancing innovation with ethical considerations amid a rapidly evolving AI industry.

Anant Bhardwaj, CEO of Instabase, shares insights on creating effective enterprise AI solutions. He underscores the importance of predictability over accuracy, arguing that user confidence hinges on reliable outcomes. The conversation compares AI agents to robotic process automation, highlighting the need for human oversight in complex workflows. Bhardwaj advocates for Retrieval-Augmented Generation to enhance AI systems and discusses the barriers to AI adoption in enterprises, stressing hands-on engagement and tailored strategies for success.

David Talby, an expert in AI governance at John Snow Labs and Pacific AI, shares insights from the 2025 AI Governance Survey involving over 350 industry professionals. He reveals that despite 30% of organizations having AI models in production, there are alarming gaps in risk management practices. The discussion spans AI adoption trends, emphasizing the dual roles of developers and deployers. Talby also stresses the importance of formal policies and unified platforms for effective governance and collaboration in navigating the evolving regulatory landscape.

Kostas Paralis, co-founder of Typedef, introduces Fenic, an innovative open-source framework for AI-driven data processing. He unpacks how Fenic treats inference as a key operation, transforming unstructured data management, and optimizing data pipelines. The conversation dives into the evolution of data technologies, the challenges of integrating structured and unstructured data, and the revolutionary potential of large language models. Kostas also discusses the importance of open file formats and practical use cases in cybersecurity and healthcare, enhancing operational efficiency.

Evangelos Simoudis from Synapse Partners shares insights on critical trends in the AI industry. He discusses the 'Great Hollowing Out,' where automation is wiping out entry-level jobs, making it harder for fresh graduates. The conversation dives into the economic implications of AI hardware depreciation and changing team dynamics that blend experienced and novice talent. Simoudis also explores the evolution of AI business models, particularly in light of GPT-5, reshaping how companies engage with AI services.