Google Launches YouTube Playables Builder
Today's curated insights on AI product management from 100+ sources across X, LinkedIn, and YouTube.
Google Launches YouTube Playables Builder
From X
AI Product Launches & Updates
-
"Your Year with ChatGPT" rollout: OpenAI @OpenAI announced the feature is now available to users in US, UK, Canada, New Zealand, and Australia with saved memory and chat history enabled, urging users to update their app.
-
YouTube Playables Builder launch: Google DeepMind @GoogleDeepMind unveiled the Playables Builder web app powered by Gemini 3, enabling creators to develop bite-sized games with text, video, or image prompts.
-
v0 + Glean integration: v0 @v0 introduced a connection to Glean workspaces, allowing teams to generate auth flows, design UI prototypes, and draft website copy directly from internal documentation.
AI Tools & Applications
-
AI SDK v6 features: Guillermo Rauch @rauchg detailed AI SDK v6 including Agents, MCP, tool improvements, structured tool calling, DevTools, reranking, JSON Schema, and image editing.
-
d0 agent for Vercel: Guillermo Rauch @rauchg shared that the d0 agent has transformed his workflow, letting him query Vercel regions by function invocation metrics through simple prompts.
-
Segment Anything Model: AI at Meta @AIatMeta highlighted how fine-tuned SAM automates real-time river mapping to support flood monitoring and disaster response with USRA and USGS partnerships.
Product Management Insights & Strategies
-
AI adoption via vibe coding: Claire Vo @clairevo shared how Brian Greenbaum at Pendo used vibe coding during parental leave to come back energized, offering a playbook for org-wide AI integration in product and design.
-
Rising skill floor & ceiling: Aakash Gupta @aakashg0 observed that AI makes technical people 10Ć more technical, emphasizing the value of PMs who can prototype by coding and engineers shipping solo.
-
Riskiest Assumption Tests: George from Prodmgmt.world @nurijanian recommended a comprehensive comparison of Riskiest Assumption Tests as a must-bookmark resource for PMs and designers.
AI Industry Developments & News
-
World models startup funding: Aakash Gupta @aakashg0 reported that Yann LeCunās AMI Labs raised ā¬500M at a $3.5B valuation to build world models from video and spatial data.
-
Intelligence debate heats up: Demis Hassabis @demishassabis argued that critics are confusing general intelligence with universal intelligence, calling the brain the most general phenomenon known.
-
Terminology clash on generality: Yann LeCun @ylecun countered that the debate is a vocabulary issue, rejecting equating āgeneralā with āhuman-levelā given the specialized nature of the human mind.
From YouTube
ChatGPT launched something insane...
Greg Isenberg ā¢ December 22, 2025
Greg Isenberg breaks down OpenAIās quiet launch of Skills for Codexāagent-skills.ioācompatible bundles that let ChatGPT and Codex call reusable instruction packsāand differentiates skills from sub-agents and MCPs, then shares a face yoga trend, his go-to Things app, a āLast 20ā expert-help startup idea, and a 6-step viral app validation framework for 2026.
Key Takeaways:
- OpenAI now officially supports Skills for Codex, enabling users to invoke reusable instruction, script, and resource bundles via $skill_name or prompt selection, using a folder with skill.md per the agent-skills.io open standard.
- Skills are standalone guides for tasks like email drafting or spreadsheet analysis, sub-agents create parallel LLM workers for functions such as code review and test writing, and MCPs grant LLMs access to external tools like Linear ticket updates or GitHub CI fixes.
- The āLast 20ā startup concept offers a marketplace where near-finished vibe-coders can book 15-minute sessions with experts for $15ā$50 (or agency subscriptions) to overcome final development roadblocks and ship their projects.
Shipmas Day 17: How I Plan To Use AI In 2026
All About AI ā¢ December 22, 2025
All About AI outlines his four strategic focuses for 2026āscaling automation, launching micro AI businesses, exploring generative AI media, and continuous learningādemonstrating both an automated news-scraping pipeline and a generative image-editing app.
Key Takeaways:
- He built a GitHub Actions workflow using Gemini 3 Flash and the Google Grounding API to scrape AI news URLs, extract headlines and content, and store them in a Supabase database on a cron schedule.
- He plans to create and A/B test multiple micro AI businesses with agentic workflows to identify scalable, automated revenue streams.
- He showcased an image editor app powered by the said image turbo inpaint model, enabling instant edits like blue lipstick and red eye shadow on uploaded images.
The AI pioneerās playbook: How to lead a transformation in your team & company
How I AI Podcast ā¢ December 22, 2025
Brian Greenbaum of Pendo shares how he initiated and scaled an AI transformation across product, design, and PM teams by leveraging a Cursorābuilt prototype, orchestrating biweekly handsāon workshops and an async Slack channel, and establishing an internal AI knowledge base with approved tools and policies.
Key Takeaways:
- On paternity leave, Brian used Cursor to prototype a QR-code music player in hours, then convinced leadership via Slack to launch an AI upskilling initiative for Pendoās product organization.
- The āProduct AIā program runs biweekly hands-on workshops (e.g., using Bolt.new to build a to-do list app) alongside an asynchronous Slack channel for āradical many-to-many sharingā of AI tools, prompts, and experiments.
- By Q1 OKRs, he co-created an internal āAI Knowledge Centerā documenting approved AI tools, data policies, and procurement paths, and measured adoption through sentiment surveysādriving significant lifts in awareness and confidence.