🔓 Google launches Gemma 4

Good Morning! Google just dropped Gemma 4, its most capable open model family yet, and for the first time, it's completely unshackled with an Apache 2.0 license. Plus, I’ll show you how to build on-brand PowerPoint presentations with Claude AI.

Plus, in today’s AI newsletter:

• Google Drops Gemma 4 with Full Apache 2.0 Licensing

• Cursor launches Cursor 3

• Alibaba Launches Qwen3.6-Plus

• How to Build On-Brand PowerPoint Decks with Claude AI

• 4 new AI tools worth trying

AI MODELS

🔓 Google Drops Gemma 4 with Full Apache 2.0 Licensing

Google has officially released the Gemma 4 family, built on the exact same technology that powers the proprietary Gemini 3. But the biggest news isn't just the performance, it's that Google has shifted from restrictive proprietary licenses to a commercially permissive Apache 2.0 license, giving developers total control over their data and infrastructure.

• The lineup includes four model sizes: Effective 2B (E2B) and 4B (E4B) for mobile and IoT devices, alongside a highly efficient 26B Mixture-of-Experts (MoE) and a 31B Dense model for workstations.

• All models natively support agentic workflows right out of the box, featuring built-in function calling, structured JSON output, and system instructions.

• The smaller edge models (E2B and E4B) are multimodal powerhouses, capable of natively processing image, video, and audio inputs locally on smartphones, Raspberry Pis, or Jetson Orin Nanos.

• Despite their highly efficient footprint, the 31B model currently ranks #3 worldwide among open models on the Arena AI Leaderboard, while the 26B MoE ranks #6, outperforming competitors up to 20 times their size.

• They are available right now on Hugging Face, Ollama, and Kaggle, and are optimized for a massive range of hardware from consumer NVIDIA/AMD GPUs to Google's own TPUs

By releasing top-tier, agentic models under Apache 2.0, Google is making a massive play for the open-source developer ecosystem. Developers no longer have to choose between cutting-edge reasoning capabilities and owning their commercial rights. Gemma 4 provides the exact recipes needed to build fully autonomous, multimodal agents on everything from a smartphone to an enterprise cluster without any restrictive licensing bottlenecks.

AI TOOLS

💻 Cursor launches Cursor 3

Startup Cursor (officially Anysphere, backed by $3 billion from Nvidia, Google, and others) just launched Cursor 3. The new release completely overhauls the platform to focus on orchestrating AI agents to automate complex programming tasks.

• Introduces a new chatbot interface where developers can describe a feature in natural language, and Cursor will generate the code alongside a demo video showing exactly how it works.

• Users can now manage and seamlessly switch between cloud-based agents (for fast, parallel processing of heavy tasks) and local desktop agents (for manual editing and testing) via a new sidebar.

• A new "Design Mode" allows developers to simply select UI elements and type out how they should be changed, with agents automatically implementing the visual tweaks.

The shift from simple code autocomplete to full-blown agentic orchestration is the next major leap in vibe coding. By letting developers manage entire teams of cloud and local agents directly within their editor, and seamlessly hand off tasks between them, Cursor is evolving from a coding assistant into a fully autonomous coding team.

AI ECONOMY

🇨🇳 Alibaba Launches Qwen3.6-Plus to Target Enterprise Revenue

Alibaba has released Qwen3.6-Plus, a proprietary AI model with a massive one million token context window. Integrated into the company's new "Wukong" enterprise AI service, the model is laser-focused on agentic coding and complex frontend development tasks.

• Qwen3.6-Plus outperforms the older Qwen3.5 model and even partially beats Anthropic's Claude 4.5 Opus in internal benchmarks.

• The model will be available via the Alibaba Cloud Model Studio API and integrated directly into the Qwen chatbot app.

• This release marks a major strategy shift for Alibaba: moving away from open-source models (like earlier Qwen versions) to monetize proprietary enterprise solutions.

• The pivot comes as Alibaba's cloud division faces intense competition from ByteDance, with Alibaba targeting $100 billion in AI revenue over the next five years.

For a long time, Alibaba was a champion of the open-source AI community. Locking down its most capable models behind an API wall signals that the honeymoon phase of free, frontier-level Chinese models might be ending. As competition with tech giants like ByteDance heats up, Alibaba is realizing that the real money lies in selling highly capable, proprietary agentic models directly to massive enterprise clients.

HOW TO AI

🗂️ How to Build On-Brand PowerPoint Decks with Claude AI

In this tutorial, you will learn how to turn Claude AI into your personal presentation assistant right inside Microsoft PowerPoint. Unlike older AI tools that generate random, messy layouts, Claude actually reads your corporate templates, layouts, and fonts to ensure every slide you generate stays perfectly on brand.

🧰 Who is This For

• Marketers creating branded presentations

• Startup founders pitching ideas fast

• Product managers sharing updates and roadmaps

• Sales teams building client-ready decks

• Consultants making polished reports

STEP 1: Install the Claude Add-in

Head over to the Microsoft Marketplace (or click "Get Add-ins" directly inside the Insert tab in PowerPoint) and search for "Claude by Anthropic." Click "Get it Now" to install the application.

Once installed, open it from your Add-ins menu, and a dedicated Claude chat window will pop up right on the side of your screen, ready to read your current slide.

STEP 2: Load Your Master Template

For Claude to truly work its magic, you must start with a properly formatted presentation. Do not just use a blank file with random text boxes. Open your template where the Slide Master, fonts, and theme colors are fully configured. Claude's true superpower is that it reads these exact layouts. Setting up a solid structural foundation ensures the AI knows exactly where the title blocks, subtitles, and image placeholders are supposed to be.

STEP 3: Prompt Your Base Slides

With your template open, tell your new AI assistant exactly what to build in the chat window.

For example, type: "Build an agenda slide with five points: Welcome, Product Launch, Meet the Team, Financials, and Q&A. Use the standard agenda layout." 

Hit enter, and sit back. Claude will automatically fetch the correct layout from your Slide Master and populate the text boxes with your bullet points in real time without you lifting a finger.

STEP 4: Automate and Scale the Deck

Now, turn the most tedious, repetitive tasks into an automated workflow.

Tell Claude: "For each of the five agenda points, create a new section header chapter slide. Use the 01, 02, 03 numbering system and add a relevant one-sentence subtitle to each." 

Claude will visually check the deck, seamlessly duplicate layouts, apply the correct numbering, and generate matching subtitles for the entire presentation. Best of all, because Claude uses your native building blocks, if you ever change your master theme colors later, the entire AI-generated deck will update instantly.

Cursor launches Cursor 3OpenAI acquires tech news show TBPN; Fidji Simo says the move aims to “help create a space for a real, constructive conversation about the changes AI creates".

Microsoft launches in-house AI models MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2, built by its superintelligence team, as it pursues “AI self-sufficiency”.

Cloudflare debuts EmDash, an MIT-licensed, TypeScript-based CMS built on Astro, designed as a serverless “spiritual successor” to WordPress, available on GitHub.

Anthropic has acquired Coefficient Bio, which was developing a platform that enables AI to run biotech tasks such as planning drug research, for ~$400M.

🧠 Trinity-Large-Thinking: Arcee AI’s powerful open model for complex, long tasks

🎥 Veo 3.1 Lite: Google’s cheaper video generation AI

🧠 Qwen3.5-Omni: Alibaba’s AI that understands text, images, audio, and video

💻 Holo 3: Open AI agent that can use computers like a human

THAT’S IT FOR TODAY

Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!

See you tomorrow :)

- Dr. Alvaro Cintas