Agentic AI

Agentic AI refers to artificial intelligence systems that can operate with varying degrees of autonomy to achieve specified goals. An AI Agent is typically composed of three core components:

• Foundation Model: A Large Language Model (LLM) or Small Language Model (SLM) that provides the core reasoning, decision-making, and orchestration capabilities. Examples include GPT-4, Claude, or smaller open-source models.

• Memory Systems: Components that maintain context through:

  • Short-term memory: Recent conversation history and immediate context
  • Long-term memory: Persistent knowledge storage using vector databases or other storage solutions

• Tools & Integrations: External capabilities that allow the agent to:

  • Retrieve information (web search, document querying)
  • Take actions (API calls, database operations)
  • Interact with other systems and services

Orchestration Tools

These are libraries used to interact with the Foundation models

Name	Team/Author	Languages	Getting Started
Instructor	Jason Liu	Python, JavaScript	Notebook
Langchain	Langchain	Python, JavaScript
LlamaIndex	LlamaIndex	Python, JavaScript
DSPy	Stanford NLP	Python
llm	Simon Willison	CLI
Amazon Bedrock	Amazon	Python
Azure AI	Microsoft	Python
Vertex AI	Google	Python
OpenAI SDK	OpenAI	Python

Agentic AI Frameworks

Name	Company	Languages
CrewAI	CrewAI	Python
LangGraph	LangChain	Python
Smolagents	Huggingface	Python
Pydantic AI	Pydantic Services	Python
LlamaIndex Workflow	LlamaIndex	Python
Phidata	Phidata	Python
Bee Agent	IBM	Python

Amazon Bedrock Agents | Amazon | Python | Vertex AI Agent Builder | Google | Python |