Скачать или смотреть Build a deep research clone with LlamaIndex workflows

In this tutorial, we'll guide you through building a "Deep Research" system using LlamaIndex, starting with basics and advancing to sophisticated agent architectures. You'll learn to create AI agents using our high-level AgentWorkflow construct, starting with single agents that can search the web and maintain context between interactions. We'll then explore creating multi-agent systems where specialized agents work together on complex tasks, with research, writing, and review agents collaborating seamlessly. Next, we'll dive deeper into our Workflows system for more fine-grained control, implementing advanced patterns like looping, branching, and concurrent execution with clear visualizations of agent behavior. Finally, we'll combine everything to build a complete Deep Research clone that breaks down topics into questions, researches answers, synthesizes findings into reports, and even uses self-reflection to improve results iteratively. By the end, you'll understand how to build sophisticated AI research systems using LlamaIndex's powerful agent and workflow capabilities.

Link to notebook:
https://colab.research.google.com/dri...

More about LlamaIndex:
https://docs.llamaindex.ai