Instagram | Sponsor this newsletter | Let’s Connect!

My fellow AI explorers

What if I told you that a small AI model, capable of improving itself, has been unleashed?

It sounds like science fiction, but it’s happening right now. This week, we’re diving deep into an innovation that could redefine AI as we know it.

Today, we’ll uncover:

• 📊 The revolutionary RAR Math model that self-evolves to master advanced math.

• 🤯 How Microsoft’s Monte Carlo Tree Search framework boosts small AI models.

• 🚀 The implications of self-improving AI for the future of machine learning.

Let’s get it!

Must See AI Tools

• 💰 Payman: AI That Pays Humans. Over 10,000+ signed up for the beta

• 💫 SubMagic: An AI tool that edits short-form content for you! (Get 10% off using code “uncoverai” at checkout)

• 🤖 Lindy: Create personalized AI agents to automate daily personal tasks and business-specific functions

• ✍️ Copy.ai: An AI-driven platform that generates high-quality marketing copy for websites, social media, emails, and advertisements.

• 🎙️ Syllaby: The only social media marketing tool you’ll ever need - powered by AI! (Get 25% off the first month or any annual plan with code “UNCOVER” at checkout)

Small AI

Can small models outsmart giants?

Every once in a while, a breakthrough changes the rules of the game.

Enter RAR Math, a small language model (SLM) developed by Microsoft that evolves through self-generated reasoning. With this approach, RAR Math surpasses even some of the largest AI models—without relying on traditional training methods.

Here’s what makes this a game-changer:

• Self-Evolution: The model iteratively improves itself by generating and evaluating its own reasoning paths.

• Monte Carlo Tree Search: A decision-making framework that allows the AI to explore possibilities and optimize problem-solving.

• Benchmark Domination: Outperforms OpenAI’s GPT-4 on specific math reasoning tasks, achieving state-of-the-art results.

RAR Math demonstrates how smaller, cost-effective models can outperform larger counterparts in tasks like the USA Math Olympiad. By leveraging Monte Carlo Tree Search, the model analyzes multiple solution paths, filters out incorrect steps, and refines its approach, creating a feedback loop that sharpens its capabilities.

This breakthrough opens doors to cheaper AI solutions, better reasoning across domains, and greater scalability for niche tasks. It’s a glimpse of a future where AI systems teach themselves and continuously evolve—without human intervention.

The future of presentations, powered by AI

Gamma is a modern alternative to slides, powered by AI. Create beautiful and engaging presentations in minutes. Try it free today.

Generative AI

Dolphin 3.0: AI on the local stage!

What if you could run an uncensored language model on your laptop?

That’s exactly what Dolphin 3.0 delivers—a lightweight model capable of unrestricted interactions.

Why you should care:

• Accessibility: No need for expensive hardware—this runs locally.

• Control: Full autonomy over your AI, without content moderation.

• Use Cases: From assumption tracking in articles to custom workflows, it’s a tool for thinkers and doers.

Running on Meta’s LLaMA architecture with optimized tweaks, Dolphin 3.0 brings uncensored creativity to the forefront. Whether you’re brainstorming, analyzing text, or experimenting with prompts, this model ensures total freedom.

30-Second AI Play

How to Create a Voice AI Agent Using 11Labs Flash 2.5

1. Set Up the Framework: Start with a platform like Retell AI, which supports AI voice integration. Configure the basic settings like compliance and latency preferences.

2. Choose a Voice Model: Select the 11Labs Flash 2.5 model for lifelike responsiveness, with just 75ms latency for real-time interactions. Adjust language preferences (English or multilingual).

3. Design Your Agent's Role: Define the agent’s purpose, tone, and key functionalities (e.g., customer support, meal plan inquiries). Use prompts with goals, instructions, and examples to tailor responses.

4. Add a Voice Profile: Pick from pre-made voices or upload a custom voice. Fine-tune parameters like speed, tone, and volume for natural conversations.

5. Test and Optimize: Run a sample dialogue. Review transcripts and adjust for conversational flow, latency, and tone to ensure a seamless user experience.

6. Integrate Phone Services: Link your agent to a phone number via platforms like Twilio. Configure settings like call duration, voicemail detection, and responsiveness.

7. Deploy and Scale: Launch the agent for customer calls. Use analytics to refine the AI further, such as adding FAQs or monitoring for sentiment trends.

✨ Pro Tip: Use this setup to automate routine calls or provide 24/7 support with a human-like touch. Perfect for businesses of all sizes!

Other Relevant AI News!

1. AI Gadgets Take Over CES 2025: From smart TVs to flying cars, this year's showcase is pure innovation. Dive deeper

2. Mississippi’s AI Push: State agencies are tasked with creating fair and secure AI policies. Read more

3. A Foldable AI Robot? Meet 'Melody,' your portable humanoid assistant. Watch here

4. AI Fighting Wildfires: Nvidia envisions robots on the frontlines of disaster relief. Read more

5. Elon Musk's Grok App: A new AI chatbot enters the race. Learn more

Golden Nuggets

• 🤖 RAR Math redefines how small models achieve state-of-the-art reasoning.

• 🐬 Dolphin 3.0 empowers local AI enthusiasts with unrestricted creativity.

• 🎰 Monte Carlo Tree Search is the unsung hero of self-improving AI.

Until our next AI rendezvous,

Anthony | Founder of Uncover AI

In partnership with