OpenAI Rolls Out ChatGPT Voice
Today's curated insights on AI product management from 100+ sources across X, LinkedIn, and YouTube.
OpenAI Rolls Out ChatGPT Voice
From X
AI Product Launches & Updates
ChatGPT Voice rollout: OpenAI @OpenAI shared you can now use ChatGPT Voice inside chat on mobile and web, enabling real-time talking, answers, and visuals.
Upgraded shopping experience on Perplexity: Arav Srinivas @AravSrinivas announced an instant checkout experience powered by memory across search and deep research modes.
Gemini 3 Pro preview in v0: v0 @v0 announced the Gemini 3 Pro preview focused on design and 3D in its platform, highlighting user-built creations.
AI Tools & Applications
Visual agent & workflow builder release: Guillermo Rauch @rauchg released a fully open source visual agent and workflow builder that outputs "use workflow" code and integrates with tools like Resend, Linear, and Slack.
LlamaSheets beta launch: LlamaIndex @llama_index announced LlamaSheets in beta to convert messy spreadsheets into AI-ready data via their LlamaCloud API.
Agent skills in DeepAgents CLI: LangChain AI @LangChainAI shared that public agent skills are now available in Deep Agents CLI, allowing agents to leverage a growing collection of capabilities.
Product Management Insights & Strategies
Perplexity release cadence: Arav Srinivas @AravSrinivas noted that since Jan 1 2025, Perplexity has shipped a new product or feature every 93 hrs and rolled out a new top model every 17 days.
Coaching vs. instructing: Lenny Rachitsky @lennysan shared that the biggest skill gap for new leaders is knowing when to coach versus telling to empower team problem-solving.
5W Framework for Product Context: Nuri Janian @nurijanian introduced a framework of WHO, WHAT, WHEN, WHERE, WHY to clarify problem context and guide product decisions.
AI Industry Developments & News
5-year AlphaFold milestone: Google DeepMind @GoogleDeepMind celebrated five years of AlphaFold 2 and highlighted usage by over 3M researchers in 190 countries.
Model leaderboard update: Aakash Gupta @aakashg0 noted that after Gemini 3 Pro’s record Elo, Anthropic’s Opus 4.5 reclaimed SWE-bench at 80.9%, showcasing three labs’ diverse strategies.
Google AI Futures Fund launch: Jeff Dean @JeffDean announced the Google AI Futures Fund + Accel Atoms, offering early access to DeepMind’s advanced models, capital, expert support, and Cloud credits.
From YouTube
Reviewing Claude Opus 4.5
Greg Isenberg • November 26, 2025
Greg Isenberg and James the Boring Marketer live-test Claude Opus 4.5 against Gemini 3 Pro by building landing pages and a clickable prototype for an “Estate Clear” probate dashboard, showcasing Claude’s one-shot design polish and how custom Claude skills streamline copy-to-code workflows.
Key Takeaways:
- Claude Opus 4.5 with the “front-end design” skill generated a production-ready HTML landing page—including hero section, social proof, and subtle animations—in a single prompt, delivering more coherent design than Gemini 3 Pro’s initial layout.
- Gemini 3 Pro in AI Studio leveraged Nano Banana Pro to first create a high-fidelity dashboard mockup image and then built the HTML around it, adding features like “polish with AI” and integrated social update functions into the prototype.
- By setting up custom Claude Code skills—an “elevated direct response” copywriting playbook based on research into Cody Sanchez’s contrarian hooks and Alex Hormoszi’s value stacking, plus the built-in front-end design skill—non-technical founders can produce conversion-optimized pages and prototypes with minimal prompts.
Deep Learning Specialization by DeepLearning.AI #BeADeepLearner
Deeplearning.ai • November 25, 2025
This video celebrates the journeys of over one million learners from diverse backgrounds who have completed the Deep Learning Specialization by Deeplearning.ai, sharing success stories and tips for overcoming common learning challenges.
Key Takeaways:
- The Deep Learning Specialization has been taken by more than one million learners worldwide.
- Participants come from varied fields—software engineering, math, physics, healthcare, and the arts—and span different life stages.
- Notable stories include a 20-year IT manager turned PhD student in reinforcement learning and a new parent who learned during naps to become a senior machine learning engineer at a startup.
How to create your own AI performance coach: Optimizing your nutrition, recovery & injury management
How I AI Podcast • November 25, 2025
Lucas Werthein demonstrates how he built a personalized AI performance coach in ChatGPT by ingesting his own MRIs, wearable metrics, blood tests, nutrition plans and training logs, then configuring prompts to balance high-performance tennis, lifting, recovery and work commitments. He shows practical examples—from meal prep for a dinner event to monitoring elbow rehab—highlighting how the coach provides data-driven, on-demand nutrition and injury management guidance.
Key Takeaways:
- Werthein aggregated multimodal personal data—MRI/X-ray images, Whoop sleep and strain CSVs, blood examinations, in-body scans, nutritionist plans and journal entries—into a custom ChatGPT to unify siloed health metrics.
- He crafted a structured prompt defining the AI as a “performance strategist and health optimization coach” with goals (“feel like a 25-year-old in a 40-year-old body”), values (precision, energy, adaptation), and hard constraints (no unproven supplements, avoid overtraining) to ensure precise, scientifically backed advice.
- Use cases include adjusting protein and carbohydrate intake before a birthday dinner, mapping a recovery timeline for an elbow injury based on daily symptom logs and PT prescriptions, and providing targeted training, mobility and sleep recommendations to extend peak athletic performance.