Hi, great question! CrewAi specializes in multi-agent system, In this tutorial I only have 1 agent, however, you can design a multi-agent system where they all talk to each other and delegate tasks. Langgraph is like a workflow that can orchestrate multiple steps. It can also be used for multi-agent systems as well by designing the workflow, how you want to control the information, where the agent can "jump" in and help. It will be a bit more complicated than CrewAi. Langchain is for injecting new information that is not on the model knowledge base. For example, if your products have complicated safety instructions, you will want the model to query your information only when it comes to safety. Or when you have multiple prompts and you want to chain them all together in the workflow (mostly for Question-answer). I personally didn't have much interaction with PydanticAI but based on their docs, they also provide AI agent framework. Because they are built on top of Pydantic, they have more of a validation output, type safety, easy for programmers to debug than other framework
Great Tutorial and very well explained. Didn't know the agent can be tasked to search for a particular page. Thank you 🙏
@@jaggyjut thank you 🥰
You are amazing.
@@AdityaSinghCodes thank you so much for always supporting!!
when to use CrewAI/Langgraph and when to use something like PydanticAI/Langchain? It's a bit confused to me at the moment
Hi, great question! CrewAi specializes in multi-agent system, In this tutorial I only have 1 agent, however, you can design a multi-agent system where they all talk to each other and delegate tasks. Langgraph is like a workflow that can orchestrate multiple steps. It can also be used for multi-agent systems as well by designing the workflow, how you want to control the information, where the agent can "jump" in and help. It will be a bit more complicated than CrewAi. Langchain is for injecting new information that is not on the model knowledge base. For example, if your products have complicated safety instructions, you will want the model to query your information only when it comes to safety. Or when you have multiple prompts and you want to chain them all together in the workflow (mostly for Question-answer). I personally didn't have much interaction with PydanticAI but based on their docs, they also provide AI agent framework. Because they are built on top of Pydantic, they have more of a validation output, type safety, easy for programmers to debug than other framework
Great Question.
Good work! 👏
Thank you so much 🎉
You nailed this tutorial. Thank you so much. Please keep them coming.
Thank you so much!!
I missed the purpose of the video and was watching you in the whole video😍
@@niksbanna thank you!
Awesome tutorial. Thank you!
@@nojepnt thank you so much!