{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Unprompted","title":"Building Predictable Conversational AI: Design, Engineering & Control | With Yam Marcovitz","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/75f4c516\"></iframe>","width":"100%","height":180,"duration":2006,"description":"Building Predictable Conversational AI: Bridging Design, Engineering & Control at Scale\nIn this episode of the CDI Expert Series: Unprompted, we sit down with Yam Marcovitz, former software architect at Microsoft and Apex.AI, and co-founder of the open-source AI framework Parlant. With a background in real-time systems and enterprise software, Yam brings a systems-level perspective to one of AI's biggest challenges: getting LLMs to consistently do what you want.The conversation explores the gap between design and engineering in conversational AI, why current tools fall short, and how open source plays a crucial role in responsible AI development. Yam shares practical insights on building agent behavior that's not just intelligent but reliable and predictable at scale.\nThe episode also explores:\nHow team collaboration and clear communication impact AI behavior as much as the technology itself\nThe limitations of current conversation design toolsSeparating design and engineering workflows to improve efficiency and clarity\nCreating flexible agents without sacrificing controlMoving beyond \"prompt and pray\" to systematic testing and validation\nManaging updates/fixing issues without retraining models from scratchDesigning for multilingual, multichannel interactions\nHow linguistic precision directly impacts agent performance\nWhy open source and transparency are key to building responsible, scalable AI systems\nThroughout the episode, Yam offers a practical, behind-the-scenes look at how robust AI systems are actually built; balancing flexibility with control, and creativity with structure. So, after reflecting on this episode: how can you build AI systems that are not just intelligent, but truly reliable, testable, and scalable?\nContact Us\nThanks for tuning in to Unprompted, for more practical lessons, and behind-the-scenes stories, exclusive events, and candid insights, visit our website or follow us on your favorite social platform.\nFollow us on social media:\nX...","thumbnail_url":"https://img.transistorcdn.com/-iQg5hMdBJ8PK4wI_rIzkGYVHTLjksUyCb3FxXpUxMI/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8yOTg2/ZWU1YjkwYTcwMWNm/NGQwZTFhNTA0Njg5/MmQ0NS5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}