OpenAI Unveils Major ChatGPT Projects Upgrade
Today's curated insights on AI product management from X/Twitter across 60+ expert sources and YouTube channels.
OpenAI Unveils Major ChatGPT Projects Upgrade
From X
AI Product Launches & Updates
ChatGPT Projects enhancements: OpenAI @OpenAI shared a Projects Update in ChatGPT adding deep research support, voice mode, improved memory for past chats, file uploads, and a mobile model selector.
Weather Lab platform: Google DeepMind @GoogleDeepMind introduced Weather Lab, an interactive AI weather model that provides real-time and historical tropical cyclone track, intensity, size, and structure predictions.
Inference endpoints milestone: Clement Delangue @ClementDelangue announced Hugging Face inference endpoints crossed 3,000 customers and cut A100 pricing to $2.50/hour.
AI Tools & Applications
999 open-source models: Jeff Dean @JeffDean highlighted Google’s release of 999 models on Hugging Face, outpacing other major AI vendors.
Top open-source agent projects: Jason Zhou @jasonzhou1993 curated the best general agent frameworks—Suna, Deer-flow, Google-gemini-search, and LangChain open deep search—for rapid prototyping.
Multi-Agent System Search (MASS): Philipp Schmid @_philschmid introduced MASS, a three-stage framework automating prompt optimization and agent collaboration for large-scale multi-agent workflows.
Product Management Insights & Strategies
GenAI Application Engineer role: Andrew Ng @AndrewYNg described the emerging GenAI Application Engineer craft that builds more powerful applications faster, as businesses hunt for this new talent.
The moral obligation of design: Lenny Rachitsky @lennysan shared Bob Baxley’s call to focus on actionable design tenets over platitudes and the “Beatles principle” for optimal team size.
Mastering AI evaluations: Teresa Torres @ttorres reflected on completing an “AI Evals for Engineers and Technical PMs” course, gaining skills in automated error analysis and data-driven product metrics.
AI Industry Developments & News
Hollywood vs. Midjourney lawsuit: Rowan Cheung @rowancheung reported major studios sued Midjourney for alleged copyright infringement, a case that could set creative AI precedents.
Perplexity Labs at GTC Paris: Aravind Srinivas @AravSrinivas covered Jensen Huang’s presentation of Perplexity Labs, showcasing new AI search and task-automation demos.
AI-driven cyclone forecasting: Demis Hassabis @demishassabis praised the collaboration between Google DeepMind, Google Research, and NWSNHC to improve cyclone prediction models and save lives.
From YouTube
Apple’s ‘AI Can’t Reason’ Claim Seen By 13M+, What You Need to Know
AI Explained • June 12, 2025
AI Explained examines Apple’s paper arguing that LLMs “can’t reason” by reviewing its tests on complex puzzles and arithmetic, highlighting methodological flaws like token-limit issues, demonstrating how tool-enabled models overcome these tasks, and offering recommendations on which models to use.
Key Takeaways:
- The Apple paper showed LLMs' performance on puzzles like Tower of Hanoi, checkers, and fox-chicken river crossing drops sharply as complexity increases, unlike deterministic software that follows explicit algorithms.
- High-digit multiplication tasks fail in zero-tool settings for models including 01 mini, 03 mini, and Gemini 2.5 Pro, but Claude for Opus autonomously invokes code execution tools to compute correct answers.
- The study ignored that some puzzles exceeded the 128,000-token output limit for Claude for Opus, causing models to truncate their responses into “shorter traces” rather than revealing new reasoning barriers.
35 years of product design wisdom from Apple, Disney, Pinterest and beyond | Bob Baxley
Lennys Podcast • June 12, 2025
Bob Baxley discusses his 35-year design journey at Apple, Pinterest, Yahoo and ThoughtSpot—defining design as a holistic mindset, explaining how to foster design-rooted cultures, and sharing frameworks like decision-driving tenets and embedding design in engineering to build better software.
Key Takeaways:
- Baxley defines design as “clear thinking made visible,” emphasizing a holistic mindset that starts with vision and mission, not just visual polish.
- He recommends embedding design under engineering as “phase zero” of product development, bringing engineers into creative discussions early to ensure feasibility and shared ownership.
- At ThoughtSpot he introduced three design tenets—“documentation is a failure state,” “start simple and let users opt into complexity,” and “the product should look like it came from a single mind”—to align distributed teams and accelerate decision-making.