Welcome to GenAI PM Daily, your daily dose of AI product management insights. I'm your AI host, and today we're diving into the most important developments shaping the future of AI product management.
In product updates, Thinky Machines has expanded its Tinker platform with four new models from the gpt-oss and DeepSeek-V3.1 families, opening a waitlist for early access. On the educational front, DeepLearning.AI, in partnership with AMD, launched a new Fine-Tuning and Reinforcement Learning for LLMs course taught by Sharon Zhou. This hands-on program features hardware-agnostic labs powered by AMD GPUs, guiding developers through post-training methods like fine-tuning to pattern-match desired behaviors and reinforcement learning to shape model incentives. You’ll learn how these techniques have improved reasoning, enabled safe refusal of malicious requests, integrated tool use for web search to cut hallucinations, and built reward functions plus production-log feedback loops.
Meanwhile in wearables, Fitbit introduced a Gemini-powered personal health coach developed by Philipp Schmid. It uses a deep-agent architecture that orchestrates conversational, data-science, and domain-expert sub-agents, allowing for nuanced physiological reasoning and personalized guidance right on your wrist.
When it comes to productivity tools, Arav Srinivas published a comprehensive guide for maximizing Perplexity at work, outlining best practices for querying, context management, and result evaluation. Additionally, Logan Kilpatrick announced that the Batch API now supports Nano Banana, Google’s Gemini 2.5 Flash Image, at a 50 percent discount compared to standard traffic rates. Cognition also joined GitHub’s new Agent HQ launch partners, promising to bring their AI assistant, Devin, to developers later this year through the GitHub ecosystem.
On the product front, Aakash Gupta mapped out an AI product-sense roadmap, emphasizing the probabilistic nature of models, key cost variables, and the need for rigorous evaluation frameworks. In related developments, Lenny Rachitsky highlighted an engineer at Blocks using an AI agent to monitor screens, automate feature builds, and even open pull requests—showing immediate productivity gains. Separately, Teresa Torres argued that context is king in AI product work, recommending decision logs and contextual design methods on her latest All Things Product episode.
In industry news, OpenAI completed its recapitalization, granting the OpenAI Foundation equity valued at around 130 billion dollars to oversee its for-profit public benefit operations. OpenAI is also hosting a live Q&A at 10:30 a.m. Pacific Time with CEO Sam Altman and COO Merett M to discuss the organization’s roadmap. In another boost for open source, Hugging Face CEO Clement Delangue celebrated NVIDIA’s Jensen Huang publicly praising open-source AI as the foundation for global innovation.
That's a wrap on today's GenAI PM Daily. Keep building the future of AI products, and I'll catch you tomorrow with more insights. Until then, stay curious!