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.
Let’s kick off with some AI product launches and updates. Arav Srinivas revealed that Perplexity will bundle its upcoming AI Excel plugin with both Perplexity Pro and Max subscription plans, expanding built-in data analysis for subscribers. DeepLearning.AI rolled out BitNet b1.58, a model version that constrains weights to –1, 0, and +1—about 1.58 bits—while sustaining competitive benchmark performance. Separately, DeepLearning.AI reported that Amazon’s Project Rainier will build an interconnected supercluster of data centers as part of a $100 billion infrastructure investment.
Moving on to AI tools and applications, Google AI demonstrated how to create an interactive fireworks display in GeminiApp using Canvas—no coding required. In related news, LlamaIndex released a fully open-source alternative to Google’s NotebookLM for completely local use, giving teams full control over their data. Another development: Claire Vo shared how she hacked together an AI-powered internal support tool for Chatprd, illustrating the benefits of customized AI solutions within organizations.
Now for product management insights and strategies. Paweł Huryn outlined four ways product discovery shifts when you’re building AI-powered offerings, from data planning to iterative validation. Another insight from Teresa Torres: she explained five types of assumptions that product teams commonly make and why systematically questioning them leads to stronger roadmaps. A quick-fire heuristic came from Lenny Rachitsky, relaying Andrew Wilkinson’s rule: if you ever ask yourself, “Should I fire this person?” even once, act on it immediately.
Finally, here are some broader AI industry developments. Additionally, Arav Srinivas proposed extending IDE-style AI features to Excel for financial analysts and invited engineers with plugin experience to collaborate. Meanwhile, Philipp Schmid released IFBench, a new instruction-following benchmark with over 1,600 test cases designed to prevent model overfitting. On a different front, Teresa Torres highlighted Tim O’Reilly’s perspective that AI isn’t ending programming—it’s fundamentally transforming how we build software.
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!