Fitbit Rolls Out Gemini Health Coach
Today's curated insights on AI product management from X/Twitter across 60+ expert sources and YouTube channels.
Fitbit Rolls Out Gemini Health Coach
From X
AI Product Launches & Updates
New models in Tinker: Thinky Machines @thinkymachines announced the addition of four new models from the gpt-oss and DeepSeek-V3.1 families to Tinker, inviting sign-ups for the waitlist.
Fine-tuning & Reinforcement Learning course launch: Andrew Ng @AndrewYNg announced a new Fine-tuning and Reinforcement Learning for LLMs course hosted on DeepLearning.AI, built in partnership with AMD and taught by Sharon Zhou.
Gemini coach in Fitbit: Philipp Schmid @_philschmid rolled out a Gemini-powered personal health coach inside Fitbit, leveraging a deep-agent architecture to orchestrate conversational, data science, and domain-expert sub-agents for complex physiological reasoning.
AI Tools & Applications
Perplexity work guide: Arav Srinivas @AravSrinivas shared a comprehensive guide on getting the best out of Perplexity for work-related use cases.
Nano Banana support: Logan Kilpatrick @OfficialLoganK announced that the Batch API now supports Nano Banana (Gemini 2.5 Flash Image) with a 50% discount versus regular traffic.
GitHub Agent HQ partnership: Cognition @cognition shared they’ve joined GitHub’s new Agent HQ launch partners, offering additional ways to try their AI assistant Devin later this year.
Product Management Insights & Strategies
AI product sense roadmap: Aakash Gupta @aakashg0 outlined a roadmap to develop AI product sense, highlighting the probabilistic nature of AI, cost variables, and the importance of building evaluation frameworks.
AI agents at work: Lenny Rachitsky @lennysan spotlighted an engineer at Blocks using an AI agent to monitor screens, automate feature development and open PRs, demonstrating AI’s immediate impact on productivity.
Context in AI product work: Teresa Torres @ttorres discussed why “context is king” in AI product work and shared practical steps—like decision logs and contextual design—on the latest All Things Product episode.
AI Industry Developments & News
OpenAI recapitalization complete: OpenAI @OpenAI announced its recapitalization, endowing the OpenAI Foundation with equity valued at ~$130B to control the for-profit public benefit corporation.
OpenAI live Q&A: OpenAI @OpenAI is hosting a live Q&A session at 10:30 am PT with Sam Altman and Merett M to discuss the future of OpenAI.
Open-source endorsement: Clement Delangue @ClementDelangue celebrated Jensen Huang’s public praise of open-source AI and Hugging Face, underlining open source as the foundation for global AI innovation.
From YouTube
Learn to align LLMs through post-training in this new course with AMD!
Deeplearning.ai • October 28, 2025
Sharon Joe, Corporate VP of AI at AMD, and Deeplearning.ai launch “Fine-tuning and Reinforcement Learning for LLMs: Intro to Post-Training,” a practical course teaching developers to align and optimize LLMs through post-training methods like fine-tuning and reinforcement learning. The curriculum covers evaluation strategies, data management, production monitoring, and hardware-agnostic labs powered by AMD GPUs.
Key Takeaways:
- Fine-tuning trains LLMs to pattern-match desired behaviors, while reinforcement learning shapes model incentives to achieve superhuman performance as critical post-training steps.
- The course details how post-training has been used to teach reasoning, tool use like web search to reduce hallucinations, and safe behaviors such as refusing malicious requests.
- Hardware-agnostic labs powered by AMD GPUs guide users through building fine-tuning pipelines, crafting effective RL reward functions, and mining production logs for a feedback data flywheel.