--- title: AI Agents description: Learn to build autonomous AI agents that can perceive, reason, and act in complex environments. Covers agent architectures, planning algorithms, multi-agent systems, and real-world applications in robotics, gaming, and automation. category: programming-tech subcategory: artificial-intelligence difficulty: intermediate, advanced url: /subject/ai-agents --- # AI Agents Learn to build autonomous AI agents that can perceive, reason, and act in complex environments. Covers agent architectures, planning algorithms, multi-agent systems, and real-world applications in robotics, gaming, and automation. ## Available Resources 1 Books • 9 Courses • 8 Websites • 5 Papers ## Courses ### 1. Multi Agent Systems Builds teams of cooperating agents with the crewAI framework, assigning roles, tools, memory, and task decomposition. Worked examples cover resume tailoring, technical article writing, customer support, outreach, event planning, and financial analysis. **Difficulty:** Advanced | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gUayts9s **Tags:** multi-agent-systems, crewai, agent-orchestration, llm-applications ### 2. Agent Design Patterns Covers core agentic patterns (reflection, tool use, planning, and multi-agent group chat) implemented in Microsoft's AutoGen framework. Projects include a two-agent conversation, a reflective blog writer, a chess-playing agent, and coding agents for financial analysis. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gzKvx5A4 **Tags:** agentic-patterns, autogen, multi-agent-systems, tool-use ### 3. Multi-Agent Use Advanced crewAI course focused on shipping agent systems: external integrations, coordinating multiple models in one crew, testing with human feedback, and deployment. Projects include project planning, a Trello progress reporter, a sales pipeline, and support analytics. **Difficulty:** Advanced | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gU9DY9kj **Tags:** multi-agent-systems, crewai, agent-deployment, workflow-automation ### 4. Computer Use with Anthropic Progresses from the Claude API through multimodal prompting, prompt caching, and tool calling to Anthropic's computer use feature, ending with an agent that reads screenshots and operates a desktop interface to complete tasks. **Difficulty:** Advanced | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gMUWg7Fa **Tags:** computer-use, claude, tool-use, multimodal, ai-agents ### 5. Evaluating AI Agents Treats agent evaluation as its own discipline: adding tracing and observability, choosing between code-based checks, LLM-as-judge, and human review, then scoring router decisions, individual skills, and full trajectories through structured experiments. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gHJtwF5s **Tags:** agent-evaluation, observability, llm-as-judge, tracing ### 6. Building Browser Agents Explains how web agents perceive pages through visual and DOM structure, then plan actions like scraping, summarizing, and form filling. Also covers AgentQ, which combines Monte Carlo tree search, self-critique, and DPO for self-correcting agents. **Difficulty:** Advanced | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gsMmCifQ **Tags:** browser-agents, web-automation, agentq, reinforcement-learning ### 7. Agent Memory Presents the MemGPT approach of treating the context window like an operating system's memory hierarchy, with core and archival tiers. Agents edit their own memory through tool calls, retaining facts and task state across long conversations. **Difficulty:** Advanced | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gNFpC542 **Tags:** agent-memory, memgpt, letta, context-management ### 8. MCP with Anthropic Covers the Model Context Protocol's client-server architecture and why it standardizes how AI apps reach tools and data. Learners build a FastMCP server exposing tools, resources, and prompts, wire a client into a chatbot, and deploy remotely. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/geffcwdq **Tags:** mcp, model-context-protocol, tool-use, claude, llm-applications ### 9. HuggingFace's Agent Course Free multi-unit course covering agent fundamentals such as tools, thoughts, actions and observations, then their implementations in smolagents, LangGraph and LlamaIndex, real-world use cases, and a graded final assignment. Requires basic Python. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gmTftTXV **Tags:** ai-agents, smolagents, langgraph, llamaindex, tool-use ## Papers ### 1. Reflexion: Language Agents with Verbal Reinforcement Learning **Author:** Noah Shinn, Federico Cassano, Edward Berman, Ashwin Gopinath, Karthik Narasimhan, Shunyu Yao Introduces a framework where an agent reflects on task feedback in natural language, stores those reflections in an episodic memory buffer, and uses them to improve on later attempts without updating model weights. Reports gains on coding benchmarks. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gGFMgjUj **Tags:** llm-agents, self-reflection, reinforcement-learning, memory, agent-evaluation ### 2. Tree of Thoughts: Deliberate Problem Solving with Large Language Models **Author:** Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Thomas L. Griffiths, Yuan Cao, Karthik Narasimhan Generalizes chain-of-thought prompting into a search over a tree of intermediate reasoning steps, letting a model consider several branches, self-evaluate them, and backtrack. Demonstrated on Game of 24, creative writing, and mini crosswords. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gRJdv_iU **Tags:** prompting, reasoning, search, llm, chain-of-thought ### 3. Toolformer: Language Models Can Teach Themselves to Use Tools **Author:** Timo Schick et al. Presents a self-supervised method by which a language model annotates its own training data with API calls to a calculator, search engine, translator or calendar, then fine-tunes on the calls that reduce perplexity, learning when to invoke external tools. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gyzrege6 **Tags:** tool-use, llm, self-supervised-learning, api-calls, fine-tuning ### 4. Generative Agents: Interactive Simulacra of Human Behavior **Author:** Joon Sung Park, Joseph C. O'Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, Michael S. Bernstein Describes twenty-five sandbox agents whose memory stream, reflection, and planning modules produce believable daily routines and emergent social behavior such as spreading invitations. Useful for understanding memory architecture and long-horizon planning in simulated multi-agent environments. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gsDCUsWm **Tags:** generative-agents, memory, multi-agent, simulation, human-behavior ### 5. ReAct: Synergizing Reasoning and Acting in Language Models **Author:** Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, Yuan Cao Interleaves reasoning traces with task-specific actions so a model can query external sources mid-reasoning, reducing hallucination on question answering and improving success on interactive benchmarks like ALFWorld and WebShop. The pattern most agent frameworks build on. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gRBH3ZRq **Tags:** llm-agents, reasoning, tool-use, react, prompting ## Websites ### 1. OpenAI's Practical Guide to Building Agents OpenAI's guide to when an agent beats a plain LLM workflow, covering model selection, tool definitions, instruction design, single- versus multi-agent orchestration patterns, and guardrails for safe production deployment. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/guRfXsFK **Tags:** ai-agents, openai, agent-design, orchestration, guardrails ### 2. Claude Code: Best Practices for Agentic Coding Anthropic's field notes on working effectively with Claude Code: structuring CLAUDE.md context files, curating tools and permissions, and running test-driven or multi-agent workflows. Readers gain concrete habits for delegating real coding tasks. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://code.claude.com/docs/en/best-practices **Tags:** claude-code, agentic-coding, ai-coding-assistants, developer-workflow, test-driven-development ### 3. Building Effective Agents by Anthropic Anthropic's engineering post arguing for simple, composable patterns over frameworks. It distinguishes workflows from agents and walks through prompt chaining, routing, parallelization, orchestrator-workers, and evaluator-optimizer designs, with guidance on when each fits. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gRWKANS4 **Tags:** ai-agents, agent-patterns, llm-workflows, anthropic, system-design ### 4. Google's Agent Companion Follow-up to Google's agents whitepaper, moving from concepts to operations: agent evaluation methods, AgentOps practices, multi-agent architectures, retrieval-augmented generation for agents, and patterns for running agent systems in production environments. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gfmCrgAH **Tags:** ai-agents, agentops, multi-agent, evaluation, google ### 5. Google's Agent Whitepaper Google's foundational paper defining what separates an agent from a standalone model, explaining the cognitive architecture of model, tools, and orchestration layer, plus extensions, functions, data stores, and reasoning frameworks like ReAct. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/gFvCfbSN **Tags:** ai-agents, agent-architecture, tool-use, react, google ### 6. Hands-On AI Engineering Curated GitHub collection of runnable Python mini-projects across AI agents, retrieval-augmented generation, OCR, and multimodal audio, each wiring real models and MCP servers into a working application you can read and adapt. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://github.com/Sumanth077/Hands-On-AI-Engineering **Tags:** ai-engineering, ai-agents, rag, mcp, python-projects ### 7. Microsoft's AI Agents for Beginners An eighteen-lesson Microsoft course with Python samples covering agent design patterns, tool use, RAG, planning, multi-agent systems, memory, and protocols like MCP and A2A, so you can assemble and deploy working agents yourself. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://github.com/microsoft/ai-agents-for-beginners **Tags:** ai-agents, agent-frameworks, mcp, multi-agent-systems, python ### 8. GenAI Agents A GitHub collection of 50-plus runnable notebooks that build agents step by step, from simple conversational bots to multi-agent research and business systems, using LangGraph, LangChain, CrewAI, AutoGen, PydanticAI and the Model Context Protocol. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://github.com/nirdiamant/GenAI_Agents **Tags:** ai-agents, langgraph, rag, multi-agent-systems, python ## Youtubes ### 1. PhiloAgents Six video lessons that build a game simulation engine where LLM agents impersonate Plato, Aristotle, and Turing. Covers agentic RAG with LangGraph, MongoDB memory, FastAPI deployment, and observability for production agent systems. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://www.youtube.com/playlist?list=PLacQJwuclt_sV-tfZmpT1Ov6jldHl30NR **Tags:** ai-agents, langgraph, rag, agent-memory, llm-apps ### 2. Building an Agent from Scratch A conference workshop that implements an agent as a plain while loop, adding planning, memory, and tool calling by hand. Watching it removes the mystery from framework abstractions and shows what an agent minimally requires. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://www.youtube.com/watch?v=xzXdLRUyjUg **Tags:** ai-agents, agent-architecture, tool-use, from-scratch ### 3. Building Agents with MCP A full workshop from Anthropic's applied AI team on the Model Context Protocol: why a single open standard replaces bespoke integrations, how servers expose tools and resources, and how to wire MCP into agents. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://www.youtube.com/watch?v=kQmXtrmQ5Zg **Tags:** mcp, ai-agents, tool-use, anthropic, integrations ### 4. Building Effective Agents Barry Zhang of Anthropic argues most teams reach for agents too early. The talk separates workflows from agents, breaks agents into environment, tools, and system prompt, and gives criteria for when each approach fits. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://www.youtube.com/watch?v=D7_ipDqhtwk **Tags:** ai-agents, agent-design, workflows, anthropic, prompt-engineering ### 5. Building and Evaluating Agents Sayash Kapoor of AI Snake Oil on why agent evaluation is hard: static benchmarks mislead, cost is rarely measured, and capability does not equal reliability. Useful before trusting any agent leaderboard. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://www.youtube.com/watch?v=d5EltXhbcfA **Tags:** ai-agents, evaluation, benchmarks, reliability ### 6. Agentic AI Overview (Stanford) A Stanford webinar framing agentic systems as a progression from plain prompting through tool use, retrieval, and autonomous planning. It gives vocabulary for comparing agent architectures and judging which level a problem actually needs. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://www.youtube.com/watch?v=kJLiOGle3Lw **Tags:** agentic-ai, ai-agents, llm, agent-architecture ## Books ### 1. AI Agents: The Definitive Guide **Author:** Nicole Koenigstein O'Reilly title covering how production agent systems are designed: planning and reactive architectures, multi-agent coordination, tool integration, performance tuning, safety strategies, and evaluation methods. Aimed at engineers moving past prototypes toward systems that stay reliable in deployment. **Difficulty:** Intermediate | **Language:** English | **Price:** Free **Link:** https://lnkd.in/dJ9wFNMD **Tags:** ai-agents, agent-design, production-ml, evaluation, deployment --- *This content is part of Dantes.io - Your Treasure Map to Knowledge* *Curated by humans at Dantes.io. Personal study use welcome; republishing this curation requires permission (team@dantes.io).* View this page online: https://dantes.io/subject/ai-agents