Amazon’s Nova 2 officially here

Hey AI Enthusiast,

Amazon just released Nova 2.

Four new AI models. Plus two game-changing services that let you build your own frontier models and deploy AI agents that actually work.

Nova 2 Lite beats Claude Haiku 4.5 on 13 out of 15 benchmarks. Nova 2 Pro competes with Claude Sonnet 4.5, GPT-5.1, and Gemini 2.5 Pro across the board.

But the real story isn't the models themselves.

It's what Amazon's letting you do with them.

Nova Forge lets you build your own customized AI models using Amazon's training data combined with yours. Nova Act delivers 90% reliability for browser automation—meaning AI agents that actually complete tasks.

This is a massive shift in how enterprises will use AI.

But first, today's prompt and community wins (then what this means for you...)

🔥 Prompt of the Day 🔥

LinkedIn AI Engagement Bot Ethics

Act as a LinkedIn growth specialist. Create one ethical AI-assisted engagement plan for [PROFESSIONAL NICHE] that builds real relationships.

Essential Details:

• Target Audience: [WHO TO ENGAGE]

• Engagement Volume: [DAILY INTERACTIONS]

• Automation Tool: [PLATFORM USED]

• Personal Touch: [HUMAN ELEMENTS]

• Relationship Goal: [DESIRED OUTCOME]

• Platform Compliance: [TOS ADHERENCE]

Create one engagement strategy including:

1. Automated vs. manual breakdown

2. Comment quality templates (10)

3. Personalization trigger points

4. Red flag avoidance rules

5. Relationship nurture sequence

6. Compliance safety measures

Grow authentically with AI assistance.

🎉 Win Wednesday 🎉

This Week's Community Wins

The wins coming in this week show exactly what's possible when you stop treating AI like a toy and start using it like a tool.

Charles Janoah Policarpio: PowerPoint in One Prompt

Charles figured out how to generate PowerPoint presentations that don't look like "AI design slop."

He's using Kimi K2 slides. The entire content and design done in one prompt.

Not a template. Not a draft. The finished presentation.

"This powerpoint one-shotted the entire content & design in one go," Charles said. "What you see is done in one prompt."

No back-and-forth. No manual cleanup. No switching between tools.

One prompt. Done.

This is what I mean when I say AI is replacing workflows, not just tasks.

Most people are still using AI to "help" them write slides. Charles eliminated the entire process.

Leigh Valentine: ICP Generator Going Into Beta

Leigh just completed the client roleplay for his Operating System and created an ICP generator.

Going into beta next week.

He's building what he calls an "emotionally attuned ICP" system. Not just demographic data. Emotional triggers, decision-making patterns, actual buyer psychology.

"I have so many plans because the things you can do with an emotionally attuned ICP is amazing," Leigh said.

This is the shift happening right now.

People aren't just using AI to automate existing processes. They're building entirely new products that couldn't exist without AI.

Leigh's ICP generator isn't a faster spreadsheet. It's a fundamentally different approach to understanding customers.

What These Wins Show

These aren't incremental improvements.

Charles didn't make PowerPoint "10% faster." He collapsed the entire process into one step.

Leigh didn't build "a better survey tool." He's creating a system that understands buyer psychology at scale.

This is what AI actually does when you use it right. It doesn't make old workflows faster. It makes old workflows obsolete.

The people winning with AI right now aren't optimizing. They're rebuilding from scratch.

Did You Know?

AI analyzing decades of lost pet reports discovered that cats and dogs use similar navigation strategies to find their way home across impossible distances.

🗞️ Breaking AI News 🗞️

Amazon Nova 2: The Full Story

Amazon announced this at AWS re:Invent on December 2nd.

Four new models in the Nova 2 family:

Nova 2 Lite - Fast, cost-effective reasoning model. Processes text, images, and videos. You can adjust how much "thinking" it does before responding. Perfect for customer service, document processing, business automation.

Beats Claude Haiku 4.5 on 13 out of 15 benchmarks. Beats GPT-5 Mini on 11 out of 17 benchmarks. Beats Gemini Flash 2.5 on 14 out of 18 benchmarks.

Nova 2 Pro - Amazon's most intelligent reasoning model. Handles text, images, video, and speech. Built for complex tasks like agentic coding, long-range planning, sophisticated problem-solving.

Competes directly with Claude Sonnet 4.5, GPT-5.1, Gemini 2.5 Pro, and Gemini 3 Pro Preview across multiple benchmarks.

Nova 2 Sonic - Speech-to-speech model for real-time conversational AI. One million token context window. Seamless switching between voice and text. Handles tasks asynchronously while you keep talking.

Integrates with Amazon Connect, Vonage, Twilio, AudioCodes, LiveKit, and Pipecat.

Nova 2 Omni - Industry first. Unified multimodal model that processes text, images, video, and speech while generating both text and images. Handles 750,000 words, hours of audio, long videos, hundred-page documents simultaneously.

Marketing teams can analyze entire product catalogs and generate complete campaigns—headlines, copy, social posts, visuals—in one workflow.

Both Nova 2 Lite and Nova 2 Pro have built-in web grounding and code execution. They can search the web for current info and run code directly.

Nova Forge: Build Your Own Frontier Models

This is the breakthrough.

Until now, you had three bad options for custom AI:

1. Customize proprietary models superficially

2. Continue training open-weight models and risk regression

3. Build from scratch at massive expense

Nova Forge changes this.

It lets you build your own optimized variants of Nova—called "Novellas"—by blending your proprietary data with Nova's capabilities.

The service pioneers "open training." You get exclusive access to pre-trained, mid-trained, and post-trained Nova model checkpoints. You can mix your data with Amazon's curated datasets at every stage of training.

The result? A customized model that combines Nova's full knowledge with deep understanding of your specific business.

You can also:

• Train AI using your own reinforcement learning "gyms"

• Create smaller, faster models through synthetic data distillation

• Implement safety controls with responsible AI toolkit

Companies using Nova Forge: Booking.com, Reddit, Sony, Nimbus Therapeutics, Nomura Research Institute.

Chris Slowe, CTO of Reddit: "We're replacing a number of different models with a single, more accurate solution that makes moderation more efficient. The ability to replace multiple specialized ML workflows with one cohesive approach marks a shift in how we implement and scale AI across Reddit."

Nova Act: AI Agents That Actually Work

Nova Act is a service for building and deploying AI agents that take actions in web browsers.

90% reliability on early customer workflows.

Powered by a custom Nova 2 Lite model trained through reinforcement learning on thousands of tasks across hundreds of simulated web environments.

Perfect for:

• Updating data in CRM systems

• Testing website functionality

• Submitting insurance claims

• Reconciling payments

• Coordinating shipments

Developers can prototype in minutes using natural language. Refine in VS Code. Deploy to AWS.

Real results:

Sola Systems - Automating hundreds of thousands of workflows per month for clients.

1Password - Users access logins with fewer manual steps across hundreds of websites.

Hertz - Accelerated software delivery 5x. Eliminated QA bottleneck. Transformed weeks of testing into hours.

Amazon Leo - Reduced weeks of engineering effort to minutes for their satellite internet launch.

Why This Matters

Amazon isn't just releasing better models.

They're giving enterprises the tools to build their own frontier AI and deploy agents that actually complete tasks reliably.

Nova Forge means you're not locked into using Amazon's models as-is. You can train your own using their infrastructure and data.

Nova Act means AI agents aren't vaporware anymore. 90% reliability is production-ready.

This is enterprise AI done right. Not demos. Not prototypes. Production systems.

Over to You...

Between OpenAI, Google, and Amazon's AI models which are you actually using in production?

Curious what's working for you.

To smarter model choices,

Jeff J. Hunter 
Founder, AI Persona Method | TheTip.ai

NEW: “AI Money Group” to Learn to Make Money with AI