The Tea ☕

☕ Nvidia announces new open AI models and tools for autonomous driving research

Source: The Verge (Dec  1, 2025)

Nvidia unveiled a new set of AI infrastructure and models aimed at powering “physical AI” — robots and autonomous vehicles that can understand and interact with the real world. The highlight is Alpamayo-R1, an open reasoning vision-language model designed specifically for autonomous driving and built on Nvidia’s Cosmos-Reason framework. By combining image and text processing with step-by-step reasoning, the model aims to give vehicles more “common sense” as they navigate complex environments, a key requirement for achieving Level 4 autonomy. Nvidia has made Alpamayo-R1 available on GitHub and Hugging Face.

To support developers, Nvidia also released the Cosmos Cookbook, a suite of guides, workflows, and resources for training and fine-tuning Cosmos models, covering areas like data curation, synthetic data creation, and evaluation. These launches underscore Nvidia’s accelerating investment in physical AI, a vision championed by CEO Jensen Huang and chief scientist Bill Dally, who see robotics as the next major frontier for AI.

☕ AWS launches new Nova AI models and a service that gives customers more control

Source: TechCrunch (Dec 2, 2025)

AWS unveiled Nova 2, a new lineup of four upgraded homegrown AI models, during CEO Matt Garman’s re:Invent keynote. The expanded Nova family now includes cost-efficient reasoning models, advanced multimodal systems capable of handling text, images, video, and speech, and a new speech-to-speech model for conversational AI. These models aim to support everything from everyday tasks to highly complex workflows, like coding and advanced agent-based reasoning.

Alongside the model updates, AWS introduced Nova Forge, a service that lets enterprise customers create custom “Novella” models based on AWS’s frontier models for $100,000 per year. Designed to solve the challenges businesses face when training models on their proprietary data, Nova Forge provides access to pre-, mid-, and post-trained versions so customers can fine-tune without losing core reasoning capabilities. Early adopters include Reddit, Sony, and Booking.com.

☕ OpenAI declares ‘code red’ as Google catches up in AI race

Source: TechCrunch (Dec 2, 2025)

OpenAI is feeling intensifying pressure in the AI race, leading CEO Sam Altman to declare a “code red” and shift the company’s focus back to strengthening ChatGPT. According to reports from The Wall Street Journal and The Information, Altman has paused several upcoming initiatives — including ads, shopping and health agents, and a new personal assistant called Pulse — so teams can prioritize improvements to ChatGPT’s speed, reliability, personalization, and ability to answer more questions. He has also instituted daily calls for teams working on the chatbot and encouraged temporary internal transfers to accelerate progress.

The memo signals a pivotal moment for OpenAI as it pours massive resources into growth while searching for a path to profitability. It also reflects a role reversal in the AI landscape: Google, once shaken by ChatGPT’s launch, is now gaining momentum with a fast-growing user base and strong performance from its latest model, Gemini 3, which has surpassed competitors on multiple benchmarks. Competitors like Anthropic are also closing in, underscoring why OpenAI is urgently refocusing on its core product.

🤖 Prompt of the Week (#POTW)

The Self-Managing Inbox Architect

This prompt creates an intelligent email triage system you can hand off to a VA, team member, or even use as guidelines for email filtering rules.


Analyze my typical email workload and create an automation blueprint:
My role: [YOUR ROLE]
Typical daily email volume: [NUMBER]
Main email categories I deal with: [LIST 4-6 TYPES]
Create a comprehensive email management system with:
1. Smart Sorting Rules
- Define 5-7 categories with:
- Category name and color code
- Keyword triggers and sender patterns
- Auto-action (archive, flag, forward, template reply)
2. Response Template Library
- Draft 6 pre-written responses for:
- Meeting requests when I'm unavailable
- Information requests I get repeatedly
- Soft nos that preserve relationships
- Quick wins (confirmations, thank yous)
3. Priority Matrix
- Create decision criteria for:
- Respond immediately (same day)
- Respond within 48 hours
- Batch process weekly
- Auto-archive/unsubscribe
4. Delegation Protocol
- For each category, specify:
- Can [TEAM MEMBER] handle this? Decision tree included
- What level of approval do they need?
- Template for their response
5. Vacation Mode Modifications
- Adapt all the above for out-of-office periods
Format as a visual flowchart-style guide that anyone covering for me can follow.

Why it works: This is an entire email operating system. Set it up once, refine it quarterly, and reclaim hours every week.

🛠️ The Toolbox

Your weekly must-try AI tools & resources

  • BRAND•E: Built for agencies and multi-brand companies who need to scale content production without writing a single prompt...while keeping each brand's authentic voice intact across every piece of content.
    Try BRAND•E FREE

  • Manus: An AI writing assistant that helps you draft, edit, and refine content with natural tone control—ideal for creators who want human-like storytelling without endless rewrites.
    Get FREE Credit Now

  • Lovable: Turns simple app ideas into fully functional products with clean UI and smart logic—no coding required. Perfect for founders and designers bringing MVPs to life fast.
    Get FREE Credit Now

💬 Community Corner

The 48-Hour Knowledge Transfer Challenge

Document your role so thoroughly that someone could step in tomorrow and keep things running at 80% effectiveness.

What You'll Create:
Video walkthrough of your typical day (15 min)
Written SOPs for your top 5 recurring tasks
Decision-making framework document
Emergency contacts and escalation path

Step-by-Step:

Session 1 (30 min): Brain Dump
Use your phone to record yourself explaining: "Here's what I do, here's why it matters, here's what would break if I stopped"
Voice-to-text transcript it (Otter.ai, Google Recorder, etc.)
Don't edit yet—just capture

Session 2 (30 min): Structure the Chaos Use this prompt:
I've recorded myself explaining my job responsibilities (transcript below). Transform this into:
1. A prioritized task list (most critical → nice to have)
2. A one-page quick-reference guide
3. Three "if-then" decision trees for common scenarios
[PASTE TRANSCRIPT]

Session 3 (30 min): Create Visual Assets
Record your screen doing your top 3 tasks (Loom is perfect)
Add these videos to a simple Notion page or Google Doc
Include timestamps: "Skip to 2:15 for the weird part"

Session 4 (30 min): Test and Refine
Have a colleague or team member review
Ask: "If I disappeared tomorrow, what's missing?"
Add those gaps

Bonus Points:
Set up a shared Slack/Teams channel just for "covering for [YOUR NAME]"
Create a "dumb questions welcome" FAQ document
Schedule this as a quarterly update (things change!)

Share Your Results: Post a screenshot of your one-page guide (redact sensitive info) and tell us: What surprised you most about documenting your work?

Start small, start now, and remember: your business should support your life, not consume it. Enjoy your holidays… for real this time.

Your AI Bestie who's grateful you're here

Catch you next Wednesday,

— Mia

Reply

Avatar

or to participate

Keep Reading