Google Unleashes Project Astra: Real Time Visual AI Hits Phones

🚨 TODAY'S AI NEWS: Google Rolls Out Project Astra – Real Time Visual AI Goes Live on Phones + More From Runway, ElevenLabs & Krea AI

Today’s AI drops are 🔥. Google rolled out its biggest real time AI update yet with Project Astra landing inside Gemini Live, while Runway, ElevenLabs, Krea AI, and academic teams pushed the boundaries of what’s possible in video, voice, and healthcare AI.

1️⃣ Google’s Project Astra Hits Gemini Live Real Time Visual AI is Here

Google is rolling out Project Astra into Gemini Live, enabling:

• Real-time conversations in multiple languages
  
• About anything you point your phone at camera or screen
  
• Starting with Pixel 9 and Samsung Galaxy S25

📌 This means Gemini can now see, hear, and speak back instantly, turning your phone into a multimodal AI assistant.

2️⃣ Google Enhances Search With Gemini + Google Lens 🔍

AI Mode in Search just leveled up:

• Now supports multimodal input, meaning users can ask complex questions about images
  
• Combines Gemini’s language reasoning with Google Lens’ visual intelligence
• Rolling out now to more Google Labs users in the U.S.

3️⃣ Runway Launches Gen-4 Turbo – Lightning Fast AI Video Creation 🎬

⚡ Gen-4 Turbo can now create 10 second videos in just 30 seconds, with:

• High consistency in visuals
  
• Strong prompt understanding
  
• Available even on the free Basic tier (125 credits)

📌 This update lowers the entry barrier to fast, quality video gen for creators.

4️⃣ ElevenLabs Releases MCP Server – AI Voice in Any Platform 📞

The MCP (Multi-Channel Prompting) Server lets platforms like Claude and Cursor:

• Add AI voice to chat or assistants
  
• Perform outbound calls
  
• Use simple text prompts to trigger speech-based workflows

📌 Think voice enabled customer service, productivity tools, or AI agents.

5️⃣ UC Berkeley + Stanford Researchers Use Test Time Training for AI Video Cartoons 🧠

Using Test Time Training (TTT):

• Added fine-tuning layers to a Transformer
  
• Created minute-long cartoon-style videos
  
• Achieved strong temporal consistency from a single shot

📌 Opens the door for AI-generated animated shorts from static prompts.

6️⃣ Krea AI Raises $83M to Build the AI Video & Image Powerhouse 🧩

New funding from Bain Capital Ventures and others will fuel Krea’s mission to:

• Unify browser-based tools for generating, editing, and customizing AI media
  
• Build the Figma for AI visuals

7️⃣ Harvard + UofT Drop MedSAM2 – Breakthrough in 3D Medical AI Segmentation 🧬

Built on Segment Anything 2.1, MedSAM2:

• Generalizes across organs, scan types, and pathologies
  
• Reduces annotation costs by 85%+
  
• Designed for 3D medical imaging and video segmentation

📌 Major upgrade for healthcare AI tooling.

🔑 Takeaways for Founders, Devs & Innovators:

• Project Astra makes phones see, hear, understand, and speak in real time
  
• Runway and Krea are pushing limits in video creation for everyone
  
• ElevenLabs + MCP could redefine voice-based AI agents
  
• MedSAM2 + TTT videos show how AI is scaling in healthcare & animation

📢 Want AI News as It Breaks?

📌 Bookmark 👉 SoftlabsGroup.com and stay ahead of the curve with daily AI news, breakdowns & enterprise-ready insights.