• unwind ai
  • Posts
  • GitHub Copilot is now Multi-Model

GitHub Copilot is now Multi-Model

PLUS: ChatGPT-style memory in Claude and Perplexity, Multi-agent framework built on OpenAI Assistants API

Today’s top AI Highlights:

  1. The only reliable agent framework built on top of OpenAI Assistants API

  2. Move over OpenAI o1 - this library lets you build LLMs that “think step-by-step”

  3. Elon Musk is doubling the already largest supercomputer to 200K GPUs

  4. GitHub’s new AI tool lets you build web apps without any coding

  5. ChatGPT-style memory to Claude, Perplexity, and other AI assistants (100% free and opensource)

& so much more!

Read time: 3 mins

AI Tutorials

AI tools are changing how we handle financial data, and building a team of AI agents that can act as financial analysts makes it even better.

This guide shows you how to set up a multi-agent financial analyst system using GPT-4o in just 20 lines of Python code. The system integrates:
A web agent for general internet research
• A finance agent for detailed financial analysis
• A team agent for coordinating between agents
working together to deliver meaningful financial insights quickly.

We are using Phidata, a framework designed for building agent-based systems to streamline the entire setup.

We share hands-on tutorials like this 2-3 times a week, designed to help you stay ahead in the world of AI. If you're serious about levelling up your AI skills and staying ahead of the curve, subscribe now and be the first to access our latest tutorials.

🎁 Bonus worth $50 💵

Share this newsletter on your social channels and tag Unwind AI (X, LinkedIn, Threads, Facebook) to get an AI resource pack worth $50 for FREE. Valid for a limited time only!

Latest Developments

Building AI applications with multiple agents can be like herding cats – they often do their own thing. Agency Swarm, a new opensource framework, gives you the tools to create and orchestrate collaborative AI agents that work together like a well-oiled machine.

Based on OpenAI's Assistants API, it lets you define roles, customize abilities, and manage communication flows so you have fine-grained control over your AI workforce. With Agency Swarm, you can build sophisticated AI apps that tackle complex tasks by leveraging the strengths of specialized agents working together.

Key Highlights:

  1. Granular Control Over Agent Communication - Agency Swarm enables you to explicitly define which agents can communicate with each other, preventing unintended interactions and ensuring a structured flow of information.

  2. Reduced Hallucinations with Type Safety - Built-in tools utilize Instructor for automatic type validation and error correction, minimizing the risk of agents generating incorrect or nonsensical outputs.

  3. No More Prompt Guesswork - Agency Swarm puts you in the driver's seat, allowing you to fully customize the prompts for each agent, ensuring they align with your specific needs and goals.

  4. Simplified Tool Integration - Easily create and integrate your own tools or leverage existing OpenAPI schemas to extend the capabilities of your agents.

  5. Quick Start - Get up and running with Agency Swarm in minutes using a simple pip install command. Comprehensive documentation and examples are available to guide you.

LLMs struggle at tasks that require complex multi-step thinking, something that OpenAI’s o1 models were released to address. But what if you could enhance your own LLM with similar "thinking" power?

LLM Reasoners is a new library to improve your LLM's reasoning abilities. Think of it as giving your model the "think-step-by-step" abilities of o1 models but with more customization. It helps LLMs tackle complex tasks by treating reasoning as a planning problem and searching for the optimal reasoning chain. With LLM Reasoners, your LLM can excel in tasks requiring logic, planning, and problem-solving, even in specialized domains. Plus, it's packed with visualization tools to help you understand exactly how your model is thinking.

Key Highlights:

  1. Flexible and Modular - Tailor the reasoning process with customizable search algorithms, reward functions, and world models, going beyond the limitations of basic CoT.

  2. Wide LLM Compatibility - Easily integrate with Huggingface models and APIs from OpenAI, Google, and Anthropic, enhancing both existing and new projects.

  3. Visualize the Reasoning - Understand your model's decision-making process with intuitive visualization tools, debugging and gaining insights that are often opaque in models like o1.

  4. Automated Evaluation - Track progress and compare different reasoning approaches with the AutoRace framework, facilitating rapid development and improvement, similar to how o1 is assessed using benchmarks.

Quick Bites

GitHub Copilot is going multi-model, bringing in Claude 3.5 Sonnet, Gemini 1.5 Pro, and OpenAI’s o1 models. OpenAI o1-preview and o1-mini are available now; Claude 3.5 Sonnet and Gemini 1.5 Pro will roll out soon. You can choose and even seamlessly switch between the models you want to use.

Here’s the most exciting announcement though. GitHub has released GitHub Spark, an AI-native tool to build micro web applications entirely in natural language, no coding at all.

  • These micro apps can integrate AI features and external data sources without requiring any management of cloud resources.

  • You start with an initial prompt, see live previews of your app as it’s built, and even see options for the UI for each of your requests.

  • It also automatically saves versions of each iteration so you can compare your app’s versions as you go. Sign up here to join the waitlist.

Elon Musk’s xAI achieved a big feat when it assembled Colossus supercomputer with 100,000 GPUs in just 122 days. Musk has confirmed that they will soon expand Colossus to 200,000 GPUs within the next 2 months, doubling the size of what was already the world's largest and most powerful supercomputer.

A small but long-awaited feature in ChatGPT! You can now search through your chat history on ChatGPT web (this was only available on the desktop app). Plus and Team users have started getting access. Enterprise and Edu users will have access in one week. Free users will start getting access throughout the next month.

Tools of the Trade

  1. Mem0 Chrome Extension: Mem0 brings ChatGPT-style memory to all your favorite AI assistants. Share context seamlessly across ChatGPT, Claude, Perplexity, and more, making your AI interactions more personalized and efficient. It is completely free and opensource.

  2. Sonnet Crawler: You can now use Claude 3.5 Sonnet to quickly scrape websites and get structured data based on your objectives. It outperforms the o1 crawler, supports batch scraping, and is monitored by AgentOpsAI for smooth operation.

  3. Polaris: Opensource data collection and evaluation framework for macOS to facilitate training and deployment of AI agents. Polaris enables screen capture, keystroke logging, and mouse tracking, all while maintaining user privacy and system performance.

  4. Awesome LLM Apps: Build awesome LLM apps using RAG to interact with data sources like GitHub, Gmail, PDFs, and YouTube videos through simple text. These apps will let you retrieve information, engage in chat, and extract insights directly from content on these platforms.

Hot Takes

  1. If LLMs were good at nothing else beyond taking unstructured data & structuring it and effective summarization & compression of text, that would be still a pretty big deal for many industries and researchers. ~
    Ethan Mollick

  2. + Stop listening to people who say in 5 years there will be no need for developers + Stop listening to people who think you can build complete products just with prompts ~
    Markus Odenthal

Meme of the Day

That’s all for today! See you tomorrow with more such AI-filled content.

🎁 Bonus worth $50 💵 

Share this newsletter on your social channels and tag Unwind AI (X, LinkedIn, Threads, Facebook) to get AI resource pack worth $50 for FREE. Valid for a limited time only!

Unwind AI - X | LinkedIn | Threads | Facebook

PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one, two (or 20) of your friends 😉 

Reply

or to participate.