✨ Welcome back, Entrepreneurs

In this issue: AI is transforming work processes, evolving from experimental tools to essential infrastructure. This includes multimodal creation, ChatGPT personalization, layered image editing, browser-native agents, and deeper integration into healthcare and real-world applications.

BIG AI MOVES

🔹 Gemini App Adds Direct Notebook Uploads

Google’s Gemini app now supports direct Notebook uploads, letting users merge projects, generate images or apps from existing work, and extend notebooks with live web research. The update positions Gemini as a flexible creation hub for building, remixing, and scaling ideas in one place.

🔹 ChatGPT Adds Controls for Tone and Style

OpenAI has introduced new personalization settings in ChatGPT that let users tune ChatGPT’s communication style adjusting warmth, enthusiasm, and emoji usage. It gives users finer control over how ChatGPT sounds, strengthening its role as a customizable assistant for writing, brainstorming, and daily work.

🔹 Google Unveils T5Gemma 2

Google launched T5Gemma 2, a multimodal, long-context encoder-decoder model with compact checkpoints for on-device use. It adds a new vision encoder, expands context windows to 128K tokens, and improves multilingual, coding, and long-form reasoning pushing Gemma beyond text-only workflows.

🧩 Qwen Unveils Image Editing with True Layers

Alibaba’s Qwen team introduced Qwen-Image-Layered, an image model that decomposes visuals into editable RGBA layers—bringing Photoshop-style control directly into AI image generation.

Highlights

• True RGBA layer decomposition

• Edit objects without affecting the rest of the image

• Resize, recolor, move, or remove elements cleanly

• Recursive, variable-depth layers for complex edits

Why it matters
Layered generation bridges AI images and real design workflows. Qwen makes AI visuals editable, consistent, and production-ready not just pretty outputs.

⚡Claude Comes to Chrome

Anthropic is piloting Claude directly inside Chrome, allowing the AI to see webpages, navigate tabs, fill forms, and debug sites while testing new safeguards against prompt-injection attacks.

Highlights

• Reads pages and interacts with live websites

• Verifies work using DOM + console access

• New permission layers reduce attack success nearly 50%

• Available to Pro, Team, Enterprise, and Max users

Why it matters:
Browser-native agents are powerful and risky. This pilot shows how AI can operate inside the web while keeping users in control.

📹 Create Videos and Ads with Pippit

In this tutorial, you’ll learn how to use CapCut’s Pippit AI agent to generate videos, photos, and full campaigns in minutes giving you a creative team on autopilot.

How it works

1. Choose a mode (ads, avatars, product photos, graphics)

2. Enter a short prompt

3. Upload assets

4. Customize layout and style

5. Export or publish instantly

Pro tip: Short, specific prompts = better results.

🩺 AIHealthTech Insider: When Algorithms Change Outcomes

Clinical AI is moving from studies into daily care. This issue covers cancer detection, ADHD treatment, dementia screening, vaccine design, glucose monitoring, and AI deployment across global health systems.

👉 Read AIHealthTech Insider: Issue #79 for the full roundup.

🚀 Sunday AI: The Week AI Agents Crossed the Line

This issue marked a shift. AI agents moved into real deployment across cloud platforms, government workflows, and enterprise tools. Inside: curated events, learning paths, and hiring signals shaping what comes next.

👉 Read Sunday Special Issue #26 for the full briefing.