WakeTheAI
Posts
🎥 AI can now create 1-minute cartoons

🎥 AI can now create 1-minute cartoons

PLUS: Google adds Multimodal Search to AI Mode

Jaynit Makwana & Dhaval Makwana
April 09, 2025

Sign up | Sponsor

Hola, AI fam 🤖

In today’s WakeTheAI edition:

AI can now create 1-minute cartoons
Google’s AI Mode gets smarter
Amazon’s new AI browser agent
Prompt: ABCDE Task Prioritization Framework
OpenAI raises $30B at a $300B valuation
Snapchat launches Sponsored AI Lenses
Deep Research is now live on Gemini 2.5 Pro
Apple brings Apple Intelligence to Vision Pro
Microsoft released an AI-generated Quake II demo
OpenAI plans to release a powerful open-weight language model

NVIDIA & STANFORD
🎥 NVIDIA & Stanford unlock 1-minute cartoons

Lazy Bits: NVIDIA and Stanford researchers developed a new AI technique called Test-Time Training, allowing models to generate full-minute cartoon videos with consistent characters, motion, and storytelling across multiple scenes.

In-Depth Details:

Cartoon-Level Demos: The system produced full one-minute Tom and Jerry-style animations, showcasing smooth transitions, dynamic motion, and narrative flow across scenes.
Memory-Driven Design: TTT layers act as learnable memory units, helping the model maintain visual and contextual continuity over extended sequences during generation.
Model Extension Technique: By adding TTT layers to existing video diffusion models, the approach scales their capabilities far beyond the original time limits.
Evaluated by Humans: In user studies, the method outperformed previous video models in consistency, motion accuracy, and overall storytelling quality.

Lazy Conclusion: Most AI videos today fall apart after a few seconds. With TTT, we’re entering an era where models can remember, adapt, and tell full stories without breaking character.

TOGETHER WITH JULIUS AI

Lazy Bits: Julius is your all-in-one AI assistant for data and dev work. It cuts hours of manual tasks down to minutes so you can spend more time building, less time fixing.

Whether you're a student, founder, or developer, Julius handles the heavy lifting across:

Data analysis & visualization: Upload files or paste data, and Julius instantly generates insights, charts, and dashboards-no spreadsheet headaches.
Code generation & debugging: Write, optimize, or fix code in seconds. Julius supports multiple languages and handles everything from scripts to SQL queries.
Math from images: Snap a photo of a math problem and let Julius solve it—perfect for students and anyone working with formulas.
File conversions: Convert files across formats like CSV, JSON, Excel, PDF, and more with zero hassle.

Why juggle 10 tools when one AI can do it all?

Start using Julius AI for free (click here)

GOOGLE
🔍 Google adds Multimodal Search to AI Mode

Lazy Bits: Google upgraded AI Mode with multimodal search, combining Gemini's vision-language capabilities with Lens to let users ask questions about images in real-time.

In-Depth Details:

Visual + Language Fusion: Users can now upload or snap a photo, ask detailed questions, and get rich, contextual answers powered by Gemini and Google Lens.
Scene-Level Understanding: AI Mode doesn’t just identify objects. It grasps full scenes, materials, and object relationships using Gemini’s multimodal reasoning and query fan-out techniques.
Search Behavior Shift: Queries in AI Mode are 2x longer than traditional search, often focused on planning, comparisons, and how-tos.
Wider Access: Initially exclusive to Google One AI Premium, AI Mode is now rolling out to millions of Labs users in the U.S. via the Google app on Android and iOS.

Lazy Conclusion: Google’s AI Mode is turning search into a full-sensory experience. It’s no longer just what you type — it’s what you see, feel, and wonder, all answered in one place.

AMAZON
🤖 Amazon’s new AI browser agent: Nova Act

Lazy Bits: Amazon introduced Nova Act, an AI model trained to perform real actions inside web browsers. A major step toward reliable, task-completing AI agents.

In-Depth Details:

Action-First AI Agent: Nova Act can complete multi-step tasks on websites. It interacts with web pages like a real user, without relying on APIs.
Benchmark Performance: Nova Act scored 93.9% on ScreenSpot Text, 87.9% on ScreenSpot Icon, and 80.5% on GroundUI. It outperformed OpenAI and Anthropic models.
Reliability Focus: The model handles tricky UI tasks like dropdowns, date pickers, and popups. It is built to reduce errors and human supervision.
Real-World Use: Nova Act powers Alexa+ to finish web tasks autonomously. It steps in when APIs cannot get the job done.
Availability: The Research Preview SDK is now live at nova.amazon.com. Developers can use it to build and test browser agents.

Lazy Conclusion: While most AI agents today need constant human supervision, Amazon’s Nova Act is pushing toward a future where AI can reliably execute digital tasks, quietly transforming how we interact with the web.

ABCDE Task Prioritization Framework

You can ask ChatGPT to act as your productivity strategist, helping you implement the ABCDE Task Prioritization Method to organize and manage your tasks effectively.

ChatGPT will analyze your task list, categorize each task based on its importance and urgency, and present a clear, structured framework.

The output will include practical tips and actionable recommendations to help you focus on high-impact tasks and optimize your productivity.

Prompt:

You are a productivity strategist specializing in the ABCDE Task Prioritization Method. Your task is to develop a clear, structured framework that organizes and prioritizes tasks effectively based on their level of importance and impact. The output should address the following key areas:

Key Focus Areas:

1. Clarifying Task Prioritization Principles – Explain the ABCDE method and how it supports effective time and task management.
2. Analyzing the Task List – Evaluate each task and categorize them into the appropriate priority level.
3. Structuring the Task Overview – Present a clear, easy-to-understand table with tasks distributed across A, B, C, D, and E categories.
4. Providing Categorization Justifications – Add a short explanation for why each task belongs to its respective category.
5. Enhancing Task Management – Offer recommendations on how to approach tasks in each category to maximize productivity.

Key Information to Include:

Task List: [List your current tasks]
Priorities: [Outline key priorities and objectives]
Available Resources: [Mention available time, tools, budget, and team capacity]
Deadlines: [Include specific due dates or urgency indicators]
Delegation Options: [Describe available delegation resources, e.g., team members, external help]

Output Requirements:

• Present the task prioritization in a markdown table format with five columns:
A (Most Important)
B (Important but Less Critical)
C (Nice to Do)
D (Delegate)
E (Eliminate)
• Include a brief explanation in each row to justify why the task was placed in that category.
• Provide additional bullet points with practical tips on how to manage and execute tasks within each category.
• Ensure the output is actionable, simple to follow, and adaptable to various work environments and priorities.

Result:

Snapchat launches Sponsored AI Lenses, letting brands create immersive, AI-generated ad experiences that boost engagement, virality, and personalized storytelling.
Deep Research is now live on Gemini 2.5 Pro, letting Gemini Advanced users generate smarter reports and audio summaries across all platforms.
Microsoft released an AI-generated Quake II demo via Copilot, showcasing game simulation tech, but admits major flaws like object memory and gameplay issues.

OpenAI raises $40B at a $300B valuation, backed by SoftBank & Microsoft. $18B will fund its Stargate project to build AI data centers across the U.S.
Apple brings Apple Intelligence to Vision Pro with visionOS 2.4, adding AI writing tools, image creation, smart search, and new spatial content experiences.
OpenAI plans to release a powerful open-weight language model with advanced reasoning in the coming months, CEO Sam Altman announced.