What is today’s beat?

• Stanford STORM - Academic Research Chat

• Hugging Face released SmolAgents

• Eliza Typescript Agent

• OpenAI tutorial

• 3 free short-courses and 3 tools for entrepreneurs

  
  
  Your FREE newsletter
   share
  to show support
  

🎯 RELEASES 🎯

Stanford
AI-Powered Academic Research Chat

Synopsis

Stanford University has introduced STORM (Synthesis of Topic Outlines through Retrieval and Multi-perspective Question Asking), an AI-driven research tool designed to generate comprehensive, Wikipedia-style articles complete with citations in mere minutes. Launched in early 2024, this open-source project aims to enhance the efficiency and accuracy of academic writing and content creation and is a great alternative to Google NotebookLM for researching topics.

Core Observations

1. Research Content Generation: STORM employs a multi-agent system where AI agents collaborate to perform content retrieval, pose multi-perspective questions, and synthesise information, resulting in well-structured and accurate articles.
   

2. Technology Stack: Powered by Bing Search and Azure OpenAI GPT-4o-mini, STORM scans, ingests, and summarises.
   

3. User Collaboration feature: Users can choose between autonomous AI operation or a collaborative human-AI interaction (Co-STORM). This means you can conduct work with colleagues.
   

4. Library: create a library, and curate knowledge - it organises and “collects” a knowledge base for you to query and learn about.

Broader Context

STORM addresses common limitations of large language models (LLMs) in academic settings, such as accuracy, specificity, and citation requirements. This is an excellent example of how agentic AI can be applied to learning by providing verified, fact-based outlines that offer multiple perspectives. Its open-source nature is true to academia, and has been gaining lots of traction online since its early 2024 release. The most powerful part of this is its library feature where you can curate a knowledge base and then query it.

HuggingFace
SmolAgents, a light barebones library for agents

Synopsis

Hugging Face has unveiled SmolAgents, a lightweight library designed to simplify the creation and deployment of AI agents. By enabling developers to implement powerful agents with minimal code, SmolAgents is a tool to call tools and orchestrate other agents.

Core Observations

1. Minimalistic Design: allows the development of AI agents in just a few lines of code.

2. Code Agent Support: The library emphasises 'Code Agents,' where agents perform actions by generating and executing Python code, enhancing flexibility and functionality.

3. Extensibility and Integration: SmolAgents supports various Large Language Models (LLMs), including those from Hugging Face Hub and integrations via LiteLLM from LLM providers like OpenAI, Anthropic, and more.

Broader Context

SmolAgents represents a significant advancement in AI development by lowering the technical barriers to creating intelligent agents. Its minimalist design and support for code-based actions enable rapid prototyping and deployment, specifically with higher order actions like orchestration. While specific benchmark results for SmolAgents are not extensively documented, the library's design focuses on efficiency and performance. Its support for code agents has been demonstrated to reduce the number of steps and Large Language Model (LLM) calls by approximately 30%.

Javascript/TypeScript
Eliza - a hot open source agent!

Synopsis

Eliza is an AI agent framework designed for developers working within the JavaScript ecosystem. Built with TypeScript, it provides a well documented framework with community support for creating and deploying AI-driven solutions. Its primary focus is to enhance development efficiency while ensuring compatibility with JavaScript-based projects. Comon’ javascript devs, let’s get into it!

It’s as easy as npm install @elizaos/core

Core Observations

1. TypeScript: Eliza is built entirely in TypeScript, encouraging devs to learn.

2. Multi-Agent Framework: Build and deploy autonomous AI agents with consistent personalities across clients like Discord, X, and Telegram. Full support for voice, text, and media interactions.

3. Advanced Capabilities: Built-in RAG memory system, document processing, media analysis, and autonomous trading capabilities. Supports multiple AI models including Llama, GPT-4, and Claude.

4. Developer-Centric Documentation: Comprehensive and clear documentation is provided to facilitate onboarding and accelerate development processes, making it accessible to developers at varying expertise levels.

Broader Context

AI devs are mostly doing it in python, but this is changing. Eliza is positioned to address the growing demand for AI tools within the JavaScript ecosystem, a widely used environment for web and software development. Its focus on type safety, modularity, and advanced capabilities (listed above, like built-in RAG) allows developers to implement AI solutions more efficiently.

Trending

• Maximize your LLM ROI - whitepaper from turing

• AiOpsLab from Microsoft - orchestration tools for infrastructure

• Alibaba slashes prices on large language models by up to 85% as China AI rivalry heats up

We are 100% free!

And with your support, we create more FREE content!

Please share us with a friend!

⚙️ BUILDERS BYTES ⚙️

What will you learn today?

We will look at OpenAI API. You will learn how to build a continuous chatbot using OpenAI's API, leveraging file-based knowledge for custom responses. This is openAI’s way of doing RAG.

Key Takeaways

1. Files: Retrieval Augmented Generation uses openAIs files to create a knowledge base in chatbot interactions.

2. Prompting: Show how to use instruction, the argument for client.beta.assistants.create to make a prompt.

3. Threads: Establish dynamic chat threads for real-time, interactive conversations. This is particularly useful if we store the conversation id in a database for retrieval at a later time.

The full tutorial is available in our newsletter repo
👉️ code.

View it in short_tutorials/openai

Do you have a product in AI and would like to contribute?
👉️ email us: [email protected] 

Is there something you’d like to see in this section?
👉️ share your feedback

Trending

• KAG: Boosting LLMs in Professional Domains via Knowledge Augmented Generation (paper)

• Using Github CoPilot to explain legacy code

• OpenCV: Computer vision - Image Captioning using ResNet and LSTM

🤩 COMMUNITY 🤩

Learning

Deeplearning.ai is Andrew Ng’s AI resource - a great place to find blogs, tutorials, and more. They recently released Reasoning with o1, so today we share a few great resources to keep you on top of your game.

• AI Agentic Design Patterns with AutoGen (featuring Penn State and Microsoft) - 1h 26min

• AI Agents in LangGraph - 1h 25 min

• Reasoning with o1 (featuring OpenAI) 1h 10 min

Talks & Events

• CES, Las Vegas, Jan 7-10

Tools

To start the new year, we feature a few tools to help busy souls - especially all you entrepreneurs out there.

• A virtual employee that scours the web, organises data, and delivers results in a spreadsheet 
  (has code)

• 1Panel - Top-Rated Web-based Linux Server Management Tool 
  (has code)

• CodeGuide - your AI companion to help you code anything

THANK YOU

Found something cool?
Want something different?

Our Mission at AlphaWise

AlphaWise strives to cultivate a vibrant and informed community of AI enthusiasts, developers, and researchers. Our goal is to share valuable insights into AI, academic research, and software that brings it to life. We focus on bringing you the most relevant content, from groundbreaking research and technical articles to expert opinions to curated community resources. 

Looking to connect with us?

We actively seek to get involved in community with events, talks, and activities. Email us at [email protected] 

Looking to promote your company, product, service, or event?