OpenAI Launches ChatGPT Advanced Voice

Today's curated insights on AI product management from X/Twitter across 60+ expert sources.

From X

AI Product Launches & Updates

Advanced Voice update in ChatGPT: OpenAI @OpenAI shared that Advanced Voice is now available to all paid ChatGPT users, making conversations more natural and effortless.
Gemini 2.5 Pro preview: Sundar Pichai @sundarpichai announced that Gemini 2.5 Pro limits have doubled from 50 to 100 queries per day and the update is live in AI Studio and Vertex AI.
Fast RAG System release: LangChainAI @LangChainAI introduced a high-performance retrieval-augmented generation implementation combining SambaNova’s DeepSeek-R1 and Qdrant, achieving a 32× memory reduction for large-scale document processing.

AI Tools & Applications

Agent Flow: no-code AI agent builder: Karan Vaidya @KaranVaidya6 shared Agent Flow, an open-source drag-and-drop platform built with LangGraph and Composio for automating complex workflows via AI agents.
LangGraph drag-and-drop UI demo: Harrison Chase @hwchase17 showcased a no-code interface for constructing LangGraph agents and workflows visually, making agent development more accessible.
LlamaBot coding assistant: LangChainAI @LangChainAI released LlamaBot, an AI agent that generates full-stack web applications via natural chat, complete with live code previews.

Product Management Insights & Strategies

Evaluations critical for AI products: Pawel Huryn @PawelHuryn noted that 85% of AI initiatives fail due to ineffective evaluation metrics like “hallucination” or “toxicity,” urging PMs to develop domain-specific Evals.
Power of structured Evals: Aakash Gupta @aakashg0 explained that a well-defined Eval framework—structured tests to measure AI performance—can differentiate high-performing AI PMs from the rest.

AI Industry Developments & News

Shift to safetensors for AI safety: Clement Delangue @ClementDelangue highlighted that adopting safetensors over pickle has been one of the biggest unsung advances in AI safety over the past year.
AI assistant doomer narrative: Yann LeCun @ylecun quipped about sensational “AI doomer” headlines—underscoring the gap between media hype and actual AI capabilities.