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 previewed GPT-5.6 Sol Ultrafast, an API-only mode delivering up to 14-times faster output. It is initially available to a select customer group as capacity expands. OpenAI also introduced Computer History for the ChatGPT Mac app, which can remember activity across apps and websites to reduce repeated context in future chats. Google DeepMind launched Gemini 3.7 Flash, positioned for coding, knowledge work, and web development.
On the developer-tools front, Cursor’s cloud agents now start three times faster through Builds: continuously prepared development environments designed for longer-running work. NVIDIA gave Cursor agents access to more than 300 skills across over 30 products. Bolt added live voice dictation, letting users describe apps conversationally. Cursor also acquired the Firetiger team to build agents that follow work into production and diagnose and fix post-deployment issues.
OpenAI shared an enterprise adoption signal: its top 10 percent of customers use plugins twice as often and skills six times as often as typical firms. Madhu Guru advised teams to consider cutting at least half their prompts after each model update, because accumulated rules, examples, and formatting constraints can degrade stronger models. Kevin Yien described senior PM judgment as turning vague feedback—such as, “this is bad, make it better”—into the right action.
Claire Vo said AI transformation frequently stalls on change management, requiring first-principles workflow redesign, internal champions such as forward-deployed engineers, and leadership alignment. Dharmesh Shah warned against “FeatureMaxx”: lower build costs do not eliminate the cognitive cost of excessive features for customers.
In related agent news, Greg Isenberg and Allie K. Miller emphasized connected business context, daily brain dumps, Slack access, clear permissions, feedback loops, and watchdog agents that identify duplication, conflicts, and improvement opportunities. Guillermo Rauch highlighted Sandbox improvements, including pre-installed default tools and full customization for coding-agent environments.
Miller described her 34-agent AI workforce, led by AI chief of staff Simon and six functional AI directors. Phoebe focuses on how to ten-times output, while Toby monitors friction, missing tool access, and repeatedly corrected outputs. The system connects documents, transcripts, Gmail, calendar, Notion, Stripe, Supabase, and GitHub. Miller has 86 dictated daily-diary entries in her personal wiki and says dictation is four times faster than typing. Sub-agents run on Haiku and Sonnet, while her software factory uses reusable primitives for login, payments, social sharing, and newsletters. Its first AI First Index product is already profitable.
Perplexity added Grok 4.6 for Pro and Max users after research-benchmark testing on performance versus cost. Databricks surpassed a seven-billion-dollar revenue run rate, with more than 80 percent year-over-year growth, and raised five billion dollars for agent products including Genie, Lakebase, and Unity AI Gateway. Vercel’s AI Gateway Production Index reported lower average token prices, with DeepSeek becoming the second-largest lab by token volume, ahead of Google.
Finally, an agent workflow compared Arsenal versus Manchester City prediction-market prices with bookmaker odds. Polymarket showed 37 percent Arsenal, 36 percent City, and 28 percent draw; Kalshi showed 37, 35, and 28. Bet365 odds normalized to roughly 38 percent Arsenal, 26 percent draw, and 36 percent City. A Codex workflow used free Polymarket and Kalshi WebSockets to monitor live prices and flag outliers, identifying Manchester City on Kalshi as worth monitoring at 34 cents or lower, with no executable expected value at current asks.
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!
GenAI PM