Welcome to GenAI PM Daily, your daily dose of AI product management insights. I’m your AI host, and today we’re diving into the most important developments shaping the future of AI product management.
OpenAI added Pinned Chats on mobile and web, letting users pin conversations via the overflow menu. Mistral AI released OCR 3 to set new accuracy and efficiency benchmarks in document intelligence. And OpenAI rolled out GPT-5.2-Codex to boost agentic coding with better performance, complex-task reliability, and scalable outputs.
Meta open-sourced Perception Encoder Audiovisual, merging audio and visual streams for state-of-the-art separation. LangChain AI introduced its Python academy course on building reasoning agents with tool and model integration. And Llama Index launched LlamaParse v2 for simpler setup, improved ingestion, and up to 50 percent cost savings.
Aakash Gupta noted Harvey reached over $100 million ARR and an $8 billion valuation with top law firms, yet sees low internal usage. Cursor now ships daily in place of traditional PM structures, treating experiments as reversible bets. And Lovable’s growth strategy creates deliberate market noise through frequent feature releases and public updates.
Pawel Huryn listed five AI evaluation traps to avoid: log real user traces, cluster domain-specific failures, design targeted scenarios, measure true positive rates on critical errors, and use clear, non-overlapping labels. Meanwhile, Carl Vellotti’s approach lets tickets expire automatically, dropping low-value items to keep roadmaps focused on high-impact work.
Anthropic revisited Project Vend by deploying Claude in a mock shop, outlined emotional support safeguards, and teamed up with the Department of Energy to apply Claude to energy, biosecurity, and research challenges. Peter Yang compared major AI assistants, noting Claude’s strength in sustained workflows with project folders and custom skills while Gemini excels at multimodal content. And Udi Menkes is hiring to build financial language models for banking services, opening new PM roles in specialist AI verticals.
In video highlights, All About AI showed a Gemini 3 prototype where three agents—Jack, Claude, and Erica—visualize inner monologues as mental images stored via a two-layer memory. Lennys Podcast featured Elena Verna on Lovable’s rise to $200 million ARR by investing 95 percent of growth in new viral loops like Shopify integrations and event credits. And Shipmas Day 13 tests found GPT-Image 1.5 is four times faster than its predecessor, delivering clear infographics, though Nano Banana Pro sometimes yields cleaner layouts.
Each of these developments offers fresh opportunities for AI product managers.
That’s a wrap on today’s GenAI PM Daily. Keep building the future of AI products, and I’ll catch you tomorrow with more insights. Until then, stay curious!