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 today, OpenAI rolled out “Your Year with ChatGPT” in the US, UK, Canada, New Zealand, and Australia, enabling saved memories and chat history—be sure to update your app. Google DeepMind launched Playables Builder, a web app powered by Gemini 3 for creating bite-sized games from text, video, or image prompts. v0 now connects with Glean workspaces, letting teams generate auth flows, design UI prototypes, and draft website copy directly from internal docs.
In related news, Guillermo Rauch detailed Vercel’s AI SDK v6 with Agents, improved tool calling, structured tool support, DevTools, reranking, JSON Schema validation, and image editing. He also revealed that the d0 agent now lets developers query regional function metrics on Vercel via simple prompts. At Meta, a fine-tuned Segment Anything Model is automating real-time river mapping to monitor floods and support disaster response, in partnership with USRA and USGS.
In strategy, Claire Vo described how Brian Greenbaum at Pendo used “vibe coding” during paternity leave to return energized and kick off org-wide AI integration in product and design. Aakash Gupta noted AI is raising the technical floor and ceiling—making teams ten times more powerful—and highlighted the value of PMs who prototype in code and solo engineers. Separately, George at Prodmgmt.world curated a comparison of Riskiest Assumption Tests, a must-bookmark resource for PMs and designers.
On the investment front, Yann LeCun’s AMI Labs raised €500 million at a $3.5 billion valuation to build world models from video and spatial data. The intelligence debate heated up as Demis Hassabis argued critics are conflating general and universal intelligence, calling the brain the most general phenomenon known. Yann LeCun responded that it’s a vocabulary issue—reserving “general” for the broadest systems, not necessarily human-level intelligence.
Turning to community insights, Greg Isenberg unpacked OpenAI’s Skills for Codex launch, compatible with the agent-skills.io standard. He explained skills as reusable instruction guides, sub-agents as parallel LLM workers, and MCPs that connect models to external tools. Greg also proposed a “Last 20” marketplace where near-finished vibe-coders book 15-minute expert sessions to solve final hurdles, and he shared a six-step framework for validating viral apps in 2026. All About AI outlined four focuses for next year—scaling automation, launching micro AI businesses, exploring generative media, and continuous learning—demonstrating a GitHub Actions pipeline using Gemini 3 Flash and Google Grounding API, plans to A/B test agentic revenue workflows, and an image editor powered by an inpaint model. On the How I AI Podcast, Brian Greenbaum of Pendo detailed his AI transformation playbook: prototyping a QR-code music player on paternity leave, launching biweekly hands-on workshops with an async Slack channel, and building an internal AI Knowledge Center with approved tools, policies, and adoption metrics.
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!