Important AI Agent Terms Every Beginner Should Know
Understanding these terms will make it much easier to learn AI Agents, Agentic AI, and Generative AI.
1. Agent
An AI system that can:
• Understand goals
• Make decisions
• Use tools
• Complete tasks autonomously
Example: An AI assistant that researches information and writes a report
2. Large Language Model (LLM)
The "brain" of an AI agent that understands and generates human language.
Examples: OpenAI GPT Models, Claude, ChatGPT, Llama
3. Prompt
The instruction given to an AI model.
Example: Summarize this PDF in 5 bullet points
4. System Prompt
A hidden instruction that defines the agent's behavior.
Example: You are an expert financial advisor. Always provide concise, evidence-based answers
5. User Prompt
The request made by the end user.
Example: Find the top 10 AI startups in India
6. Context
The information the AI uses to answer a question.
Examples: Conversation history, uploaded files, retrieved documents
7. Context Window
The maximum amount of information an AI model can process at one time
8. Token
The smallest unit of text processed by an AI model.
Tokens can represent words, parts of words, punctuation
9. Embeddings
Numerical representations of text used to measure semantic similarity
Common uses: Semantic search, recommendations, RAG systems
10. Vector Database
A database optimized for storing and searching embeddings
Popular options: Pinecone, ChromaDB, Weaviate
11. Memory
Allows an AI agent to remember information
Types: Short-term memory, long-term memory, episodic memory, semantic memory
12. Tool
An external capability the AI agent can use
Examples: Calculator, web search, Python execution, email API, database
13. Tool Calling
The process of selecting and using the right tool to complete a task
14. Function Calling
A structured way for an AI model to invoke predefined functions or APIs
Example: Get weather, send email, book a meeting
15. Planning
Breaking a large goal into smaller, manageable tasks
Example: Build Website → Design UI → Develop Backend → Test → Deploy
16. Task Decomposition
Splitting a complex problem into multiple subtasks
17. Reasoning
The AI's ability to analyze information and decide the next best action
18. Chain of Thought (CoT)
A prompting technique where the model reasons through a problem step by step before answering
19. ReAct (Reason + Act)
A framework where the agent:
1. Thinks
2. Takes an action
3. Observes the result
4. Continues until the task is complete
20. Reflection
The agent evaluates its own output and improves it if necessary
21. Observation
The result received after executing an action
Example: Search Tool → Returns 5 articles → Observation
22. Goal
The final objective the agent is trying to achieve
Example: Generate a complete business report
23. Workflow
The sequence of steps an agent follows to complete a task
24. Orchestration
Managing and coordinating multiple tools, models, or agents in a workflow
