Google, Cassava Enable Data-Free GeminiApp Access
Today's curated insights on AI product management from X/Twitter across 60+ expert sources and YouTube channels.
Google, Cassava Enable Data-Free GeminiApp Access
From X
AI Product Launches & Updates
Partnership enabling data-free access to Google AI services: Josh Woodward @joshwoodward announced a partnership with Cassava Technologies at AfricaCom offering data-free access to the GeminiApp and a 6-month extended trial of Google AI Plus, featuring Gemini 2.5 Pro, 200 GB storage, NotebookLM, and Flow by Google.
New multi-agent systems course launch: Andrew Ng @AndrewYNg unveiled a new course, “Design, Develop, and Deploy Multi-Agent Systems” in collaboration with CrewAI Inc, teaching how to build teams of AI agents for complex, end-to-end workflows.
Company Research agent release: Dharmesh Shah @dharmesh launched the Company Research agent, enabling easy lookup of startups and other companies with a simple agent interface.
AI Tools & Applications
Deep agents for long-running workflows: LangChainAI @LangChainAI introduced the concept of deep agents—architected differently with four characteristics to maintain context and plan ahead for complex, multi-step tasks.
Vibe coding tips for Gemini & Google AI Studio: Philipp Schmid @_philschmid shared 9 practical prompt engineering tips for building with GoogleAI Studio and Gemini, emphasizing specificity, iterative refinement, and structured input.
8Ă— throughput in document processing: Llama Index @llama_index highlighted how Pathwork scaled life insurance document processing from 5,000 to 40,000 pages per week using LlamaParse, achieving 8Ă— improved throughput for complex medical records and scanned PDFs.
Product Management Insights & Strategies
Build speed vs distribution challenge: Lenny Rachitsky @lennysan noted that AI speeds up product development, but since competitors share that speed, PMs must focus on distribution strategies as traditional growth channels become less effective.
Avoid building for PMs as primary users: Jason Zhou @jasonzhou1993 argued that PMs lack budgets and standard processes, making them a hard-to-sell-to segment without a clear case study of a PM-focused product taking off in the last five years.
Reality check on PM handoffs: George Nurijanian @nurijanian highlighted the misalignment in product workflows, where design mocks and engineering estimates rarely follow the planned sequence, urging PMs to scrutinize and refine team processes.
AI Industry Developments & News
Fireflies AI founders' early strategy: Rowan Cheung @rowancheung revealed that the founders of Fireflies AI pretended to be an AI notetaker by joining meetings on mute and manually taking notes before AI tech existed, a tactic that contributed to its $1B+ valuation.
Balancing awe and fear of AI: Mustafa Suleyman @mustafasuleyman emphasized that understanding AI fully requires both amazement at its capabilities and fear of its risks, underscoring the dual-edge nature of AI advancement.
Google's largest AI investment in Germany: Sundar Pichai @sundarpichai announced Google's biggest investment in Germany to date, reinforcing its commitment to help businesses and citizens succeed in the AI era.
From YouTube
Design, Develop, and Deploy Multi-Agent Systems with CrewAI
Deeplearning.ai • November 11, 2025
João Moura introduces Deeplearning.ai’s new course "Design, Develop, and Deploy Multi-Agent Systems with CrewAI," covering end-to-end guidance on building scalable, trustworthy AI agent systems using CrewAI’s open-source framework.
Key Takeaways:
- JoĂŁo Moura highlights the transformative potential of AI agents to automate fraud detection in finance, product returns in retail, and customer behavior analysis in communications.
- The course covers CrewAI’s open-source framework fundamentals, including agent/task definitions, inter-agent communication patterns, tool integration (e.g., web search or MCP), and memory architectures with guard rails and flows.
- To prepare agents for large-scale deployment, the video presents production best practices such as tracing for error source identification, hooks for execution lifecycle management, and evaluation methods for ongoing performance monitoring.