Gemini API Delivers 3× Faster Video Responses
Today's curated insights on AI product management from X/Twitter across 60+ expert sources and YouTube channels.
Gemini API Delivers 3× Faster Video Responses
From X
AI Product Launches & Updates
Endpoint Update for Gemini 2.5 Pro Model: Logan Kilpatrick @OfficialLoganK announced the redirection of gemini-2.5-pro-05-06 and gemini-2.5-pro-03-25 endpoints to the GA gemini-2.5-pro model on June 26 at 10 AM PST.
Hugging Face Inference Providers Auto Selection: Julien Chaumond @julien_c shared that you can now use provider='auto' to let Hugging Face automatically pick the best inference provider based on your preferred order.
Gemini API Video & PDF Performance Boost: Philipp Schmid @_philschmid announced that the Gemini API now delivers 3× faster video responses and 4× faster cached PDF processing, plus bulk video support and near-instant implicit caching.
AI Tools & Applications
Cursor AI Integration with Hugging Face: Clément Delangue @ClementDelangue announced that you can now connect Hugging Face to Cursor AI to discover models, datasets, papers, and apps seamlessly directly from the editor.
Price Movement Tracking in Perplexity Finance: Aravind Srinivas @AravSrinivas rolled out a timeline view to track price movements on any ticker in Perplexity Finance, enabling real-time financial insights powered by AI.
Context Engineering Defined: Harrison Chase @hwchase17 provided a working definition of context engineering as "building dynamic systems to provide the right context to AI models in real time."
Product Management Insights & Strategies
Five Types of Product Managers: Lenny Rachitsky @lennysan outlined five distinct PM roles—consumer PM, platform PM, growth PM, AI PM, and data PM—detailing what differentiates each focus area.
Managing Preview Model Churn: Logan Kilpatrick @OfficialLoganK emphasized reducing preview model iterations to minimize customer disruption and improve clarity when shipping updates.
Best Practices for Multi-Agent Systems: Paweł Huryn @PawelHuryn shared key guidelines like starting with stronger models for accuracy, using unified APIs via OpenRouter, and separating agent roles to streamline multi-agent workflows.
AI Industry Developments & News
Daily AI Market Roundup: Rowan Cheung @rowancheung summarized key industry moves—from Meta’s Oakley glasses partnership to Mistral’s releases, Anthropic’s safety sims, and xAI’s Grok roadmap.
AI’s Monetization Potential: Logan Kilpatrick @OfficialLoganK declared that AI is likely to be the most monetizable asset in human history, underscoring the technology’s fundamental business impact.
OECD AI Market Report Powered by TAAFT: There’s An AI For That @theresanaiforit highlighted the landmark OECD study that uses hard data to map global AI market trends, moving beyond theory to actionable insights.
From YouTube
I built next gen AI Agents with MCP + Claude (STEAL my workflow)
Greg Isenberg • June 23, 2025
Greg Isenberg and Riley Brown demonstrate how to use Docker’s MCP toolkit to turn Claude Sonnet 4/Opus 4 into an AI agent with access to external tools—walking through setting up Notion for content workflow automation and using Glyph.app to build and invoke custom thumbnail ideator workflows.
Key Takeaways:
- Docker’s MCP toolkit includes 116 pre-built tools (e.g., Notion, Firecrawl) that can be enabled via Docker and require restarting the MCP server in Claude whenever configurations change.
- The Notion MCP integration lets Claude read from and write to Notion databases—example: generating three short-form hooks in a “how to create content” database and adding them back as new entries or comments.
- Glyph.app’s public workflow builder can be called from Claude via MCP to run multi-step automations like a “thumbnail ideator” that ingests example PDFs and outputs five thumbnail designs on a canvas for quick ideation.
Build a Business In 1 Day With Claude Code and Google Veo 3
All About AI • June 23, 2025
In this video, All About AI builds ‘Zero Day Hustle’, a $2.99/month subscription AI side-hustle platform, in a single day by using Gemini 2.5 Pro for concept and branding, Claude Code to generate and deploy a Stripe-integrated landing page from a GitHub template on Versel, and Google Veo 3 for a rapid AI marketing video.
Key Takeaways:
- The Zero Day Hustle concept, named via Gemini 2.5 Pro, offers exclusive AI side-hustle video drops for $2.99/month through a Stripe-integrated landing page.
- Claude Code read the project directory, created a step-by-step development plan, generated front-end and back-end code including Stripe subscription logic, and deployed the site on Versel.
- Google Veo 3 produced a 20–30-minute AI marketing ad by scripting scenes, assembling selected AI-generated clips, and adding music from Suno to promote the platform on short-form channels.
How a 91-year-old vibe coded a complex church event management system using Claude and Replit
How I AI Podcast • June 23, 2025
John Blackman, a 91-year-old retired electrical engineer, describes how he used Claude to draft requirements and user stories and then leveraged Replit’s AI agents to build a multi-tenant church event management system—complete with QR-code registration, “impact passports,” VIN-based oil filter lookups, and automated reports—in just a few days.
Key Takeaways:
- Blackman drafted a detailed roadmap and user stories in Claude, instructing it to ask follow-up questions and produce requirements for a Replit-based application.
- The resulting app supports system and church-level admins, generates printable QR-code “impact passports,” captures digital waivers and signatures, and produces Excel and PDF reports for demographics, service usage, food orders and oil changes via a VIN lookup using the OpenAI API.
- He deployed the app on Replit for about $350, overcame production vs. development PDF-attachment issues by adjusting file formats, and credits this AI-powered process with extending his productivity into his 90s.