Chris Smith is a longtime engineering leader who has been in the trenches of building with AI & machine learning for years. He’s led the development of data systems & strategies at tech giants like early Google, Yahoo, and Sun; S&P 500's like Live Nation; and a wide variety of startups. — Topics discussed: (00:00) AI industry at inflection point, causing chaos (09:05) Machine learning, neural nets, and generative AI (14:03) Generative AI: LLMs + broad understanding (21:56) Open source models improve specialized problem solving (25:06) Access to data leads to competitive advantage (32:53) AI training improves productivity and learning speed (42:51) Reduced investment in GPT models speeds results (48:47) Expectation mismatch leads to brand perception risks (53:54) Non-technical work is crucial for AI product success (57:30) Building a computer vision product from scratch (01:03:14) A strategic approach to refining and testing prototypes (01:08:04) Closing learning loops — Links & resources mentioned Find the full transcript at: https://podcast.makethingsthatmatter.com/chris-smith-how-to-add-ai-to-product/#transcript Send episode feedback on Twitter @askotzko , or via email Chris Smith: • LinkedIn • X / Twitter: @xcbsmith • Bluesky @xcbsmith — Related episodes: • #75 Chris Smith: Simple guidelines for AI investment sizing — People & orgs: • Dr. Marily Nika - AI Lead, Meta Reality Lab • Travis Corrigan - Head of Product, Smith.AI — Books: • Evidence Guided - Itamar Gilad — Other resources: • GPT = “generative pre-trained transformer” • Wizard of Oz experiment • Tom Chi - learning loop • Joel Spolsky: The iceberg secret, revealed • ML Ops • Computer vision • Precision-Recall curves • Leaked Google memo: “There is no moat” • Universal basic income (UBI) • Stop-loss order This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit blog.makethingsthatmatter.com