- https://github.com/sunnysavita10/Agentic-2.0
- https://python.langchain.com/docs/integrations/tools/
- https://langchain-ai.github.io/langgraph/concepts/why-langgraph/
- https://www.youtube.com/watch?v=qAF1NjEVHhY&t=2s
- https://www.youtube.com/watch?v=1bUy-1hGZpI
- Concepts covered
- any model
- any embedding
- prompting
- chaining
- vector DB
- RAG
- Multi Model RAG
- parser
- Agent
- Use langraph for agentic flow.
- Agent is an advanced AI assistant that can perform complex tasks by chaining together multiple models and tools.
- Agent is an autonomous system containing
- Agentic Workflow (orchestrator creates the workflow)
graph LR
A(User Input) --> B(LLM)
B --> C(Think)
C --> D(Action - Tool)
D --> E(Observation - validate)
E --> C
E --> F(Output)
- Agentic AI - orchestration of multiple agents.
- Tools using langchain
- standard
- custom
- Langraph is a framework for building agentic flows in AI applications. It works as an orchestrator. Orchestrator is used to create agentic workflow.
- Node
- Edges
- State
- Conditional edges
Langraph
- graph()
- state()
- edges - connectivity between the functions.
- node - function
- invoke
- stategraph()
- conditional edges.
- Supervisor node -
- It is a special node that can be used to supervise the execution of the graph.
- It can be used to monitor the execution of the graph and take action if something goes wrong.
- It can also be used to log the execution of the graph.
graph LR
A([start]) --> B(supervisor)
B --> C(LLM)
B --> E(RAG)
C --> D(end)
E --> D