Welcome back to Daily AI Skills.

Here’s what we are covering today:
1. Stanford Research Paper on Longer AI Videos
2. Google’s Agent2Agent Platform
3. Big AI Announcements at Google Cloud Next 2025

+ Google’s Prompt Engineering Guidebook Document

NVIDIA, Stanford Enable Longer AI Videos

Researchers from NVIDIA and Stanford have introduced Test-Time Training (TTT), a new AI technique that pushes the boundaries of video generation, enabling minute-long animated clips with improved coherence and storytelling.

Key highlights:

• TTT enables the creation of full-minute animations with consistent visuals and narrative flow, outperforming previous methods in user studies.

• It works by adding TTT layers that function like neural memory, helping models retain context and visual consistency across extended sequences.

• The team showcased the method using Tom and Jerry-style animations, featuring dynamic scenes and smooth character interactions.

• Rather than building models from scratch, this technique enhances existing video models by integrating TTT layers, allowing them to handle longer videos than before.t May, with the feature debuting last month for Advanced subscribers.

Read the full research paper here

Google Launches Agent2Agent Protocol to Enable Cross-Platform AI Collaboration

Google has introduced Agent2Agent (A2A), an open protocol that lets AI agents from different developers and platforms collaborate seamlessly. The initiative is supported by over 50 major companies, including Salesforce, SAP, and PayPal.

Key points:

• A2A allows agents to discover each other’s skills, coordinate tasks, and share information across systems—even without a shared memory or context.

• It’s designed to work alongside Anthropic’s MCP, with A2A focusing on agent-to-agent communication and MCP handling interactions with external tools.

• Launch collaborators include major enterprises like Atlassian, ServiceNow, Workday, and top consulting firms like Accenture, Deloitte, and McKinsey.

• A2A enables complex, multi-agent workflows—such as automated hiring processes—where agents can handle tasks like sourcing candidates and running background checks with minimal or no human input.

Read the full blog here

Big AI Announcements at Google Cloud Next 2025

At its Google Cloud Next 2025 event, Google unveiled a wave of AI updates, including a new agent-powered coding platform, advanced AI chips, and enhancements to its video, audio, and image models—plus a streamlined Gemini model.

Here are the highlights:

• Project IDX is joining forces with Firebase Studio to form an agentic app development platform aimed at competing with tools like Replit and Cursor.

• Ironwood, Google’s most advanced AI chip to date, brings major gains in speed and energy efficiency.

• Model improvements span several areas: Veo 2 now supports video editing and camera control, Lyria introduces text-to-music generation, and Imagen 3 boosts image creation and editing.

• Gemini 2.5 Flash was introduced as a lightweight, faster, and more cost-efficient version of Gemini, offering adjustable reasoning capabilities for budget-conscious tasks.

Read the full announcement by Google here

Google Prompt Engineering Guidebook Document

📩Forward it to people you know who are keeping pace with the changing AI world, and stay tuned for the next edition to stay ahead of the curve!

Maybe you missed out on these:

1. Copilot vs Claude vs Gemini

2. How I Built an AI Influencer in 2 Minutes—and Landed Sponsorships

3. Microsoft Muse, Helix and Google's AI Co-Scientist

4. Bolt vs Lovable

5. Your AI Stack Just Got Superpowers: 4 New Inference Providers Live!