Agents

Agents are AI programs where a language model controls the flow of execution. The core of an Agent is a model using tools in a loop, guided by instructions:

Model: controls the flow of execution. It decides whether to reason, act or respond.
Instructions: program the Agent, teaching it how to use tools and respond.
Tools: enable an Agent to take actions and interact with external systems.

Agents also have memory, knowledge, storage and the ability to reason:

Memory: gives Agents the ability to store and recall information from previous interactions, allowing them to learn and improve their responses.
Storage: is used by Agents to save session history and state in a database. Model APIs are stateless and storage makes Agents stateful, enabling multi-turn conversations.
Knowledge: is domain-specific information the Agent can search at runtime to provide better responses (RAG). Knowledge is stored in a vector database and this search at runtime pattern is known as Agentic RAG or Agentic Search.
Reasoning: enables Agents to “think” before responding and “analyze” the results of their actions before responding, this improves reliability and quality of responses.