Resources

Agents

1. Agentic Tools & Frameworks

Latest Framework Updates (2025):

• OpenAI Agents SDK (March 2025): Minimalist Python framework for multi-agent workflows with robust tracing
• LangChain v1.0 (2025): Standard tool calling architecture, provider-agnostic design
• LangGraph v1.0 (2025): Node-based architecture for multi-agent systems with structured state management
• Anthropic Claude Skills: New API for building agents with Claude
• Model Context Protocol (MCP): Standard protocol for connecting AI models to external data sources and tools

2. Agents System Design

• Agent Design Patterns
  • Building Effective Agents Cookbook (Anthropic)
  • Anthropic Cookbook (Agents patterns)
  • OpenAI Cookbook (agent orchestration)
  • AutoGen multi-agent design patterns
• Core Concepts & Architecture
  • The Anatomy of Agentic AI
  • Llamaindex agents workflow basic
  • Dapr Agentic Cloud Ascent Design Pattern
  • Building A Generative AI Platform by Chip Huyen
  • Compound AI Systems by BAIR
• Prompting & Development
  • GPT-4 Prompting Guide

3. Learning Resources

• Comprehensive Courses
  • Deeplearning.AI Agents Courses
  • Huggingface Agents Course (GitHub)
  • Microsoft AI Agents for Beginners
  • LLM Course

3.5. Evaluation & Testing Tools (2025)

• Evaluation Platforms
  • Maxim AI: End-to-end evaluation platform for AI agents
  • Langfuse: Open-source observability and evaluation for LLM applications
  • Comet Opik: AI agent evaluation and monitoring
  • Arize: ML observability platform with agent evaluation
  • Evidently AI: AI agent benchmarks and testing frameworks
• Benchmarks
  • Web Agent Benchmarks: Evaluate agents' web navigation and interaction capabilities
  • OS Agent Benchmarks: Test agents in operating system environments
  • Tool-Using Agent Benchmarks: Assess agents' proficiency with external tools
  • Multi-Agent Benchmarks: Evaluate collaborative agent systems

4. Applications & Examples

• Awesome LLM Apps - Curated collection of LLM applications
  • Starter AI Agents

5. Additional Resources

• The Rise of Agentic AI - Comprehensive slide deck
• From LLMs to Stateful Long-Running Multi-Agent Systems

6. Recent Developments (2025)

• Computer-Using Agents (CUA)
  • OpenAI's Operator: Computer-Using Agent powered by GPT-4o
  • OpenCUA: Open-source framework for computer-use agents (August 2025)
  • Microsoft Fara-7B: Efficient agentic model for computer use
• Agentic RAG
  • Traditional RAG vs. Agentic RAG (NVIDIA): Why AI agents need dynamic knowledge
  • Uber's Enhanced Agentic RAG: Production-scale implementation
• Research Papers
  • ARAG: Agentic Retrieval Augmented Generation: Research showing 42.1% improvement over standard RAG
  • OpenCUA: Open Foundations for Computer-Use Agents: Framework for transparent research