LlamaIndex launches open-source NotebookLM

AI Product Launches & Updates

• Perplexity bundling: Arav Srinivas @AravSrinivas revealed the upcoming AI Excel plugin will be bundled with Perplexity Pro and Max subscription plans.

• BitNet b1.58 update: DeepLearning.AI @DeepLearningAI detailed BitNet b1.58, constraining weights to –1/0/+1 (~1.58 bits) while sustaining competitive benchmark performance.

• Project Rainier supercluster: DeepLearning.AI @DeepLearningAI reported Amazon’s Project Rainier will build an interconnected “supercluster” of data centers as part of a $100 billion spend.

AI Tools & Applications

• Interactive fireworks with Gemini: Google AI @GoogleAI demonstrated building an interactive fireworks display using Canvas in GeminiApp without coding.

• Open-source NotebookLM alternative: LlamaIndex @llama_index launched a fully open-source version of NotebookLM for local use.

• Custom AI support tool: Claire Vo @clairevo shared hacking an AI-powered internal support tool for chatprd, highlighting AI-driven customization benefits.

Product Management Insights & Strategies

• AI-driven product discovery: Paweł Huryn @PawelHuryn outlined 4 ways product discovery differs for AI-powered products.

• Challenging team assumptions: Teresa Torres @ttorres explained the 5 types of assumptions product teams make and why questioning them drives better products.

• Quick-fire hiring heuristic: Lenny Rachitsky @lennysan shared Andrew Wilkinson’s rule: fire immediately if you think “Should I fire this person?” once.

AI Industry Developments & News

• Excel AI opportunity: Arav Srinivas @AravSrinivas proposed extending IDE-style AI features to Excel for financial analysts, inviting plugin-experienced engineers to collaborate.

• Instruction benchmark IFBench: Philipp Schmid @_philschmid released IFBench with over 1,600 test cases to challenge instruction-following models amid concerns of overfitting.

• The future of programming: Teresa Torres @ttorres shared Tim O’Reilly’s piece on how AI is not ending programming but fundamentally transforming it.