Terms: Antigenic Agents 5/21/25

Question 1

What is an AI Agent?

Answer 1

Software program that perceives its environment, reasons, and acts autonomously to achieve goals, often using LLMs.

AI agents can perform tasks without direct human intervention and are capable of learning from their experiences.

Question 2

Define Agentic AI.

Answer 2

Field focused on creating autonomous AI agents capable of complex, multi-step tasks with minimal human intervention.

This area emphasizes the development of sophisticated AI systems that can operate independently.

Question 3

What is an Agentic Workflow?

Answer 3

System where one or more AI agents collaborate to manage, decide, and execute complex processes.

This approach allows for more efficient handling of tasks by leveraging multiple AI systems.

Question 4

What characterizes an Autonomous Agent?

Answer 4

AI agent operating independently without direct human supervision for extended periods.

Autonomous agents are designed to function in dynamic environments where human oversight is impractical.

Question 5

What is a Large Language Model (LLM)?

Answer 5

Advanced AI model trained on vast text data, used for understanding, generating, and manipulating language; often the ‘brain’ of an AI agent.

LLMs are crucial for enabling natural language processing capabilities in AI applications.

Question 6

Define Prompt Engineering.

Answer 6

Designing and optimizing input prompts for LLMs to get desired responses or behaviors.

Effective prompt engineering can significantly enhance the performance of LLMs.

Question 7

What does Tool Use / Function Calling mean in AI?

Answer 7

Ability of an AI agent to interact with external systems, APIs, or functions to perform tasks beyond the LLM’s inherent knowledge.

This capability expands the range of tasks an AI agent can accomplish.

Question 8

What is Memory in AI Agents?

Answer 8

Ability of an AI agent to store and retrieve information from past interactions. Includes short-term and long-term memory.

Memory allows AI agents to learn from previous experiences and improve their responses.

Question 9

What is Planning in AI Agents?

Answer 9

Process where an AI agent breaks down a complex goal into a sequence of steps.

Planning is essential for achieving multi-step tasks effectively.

Question 10

Define Perception in AI Agents.

Answer 10

Ability of an AI agent to gather and interpret information from its environment.

Perception enables AI agents to respond appropriately to changes in their surroundings.

Question 11

What is Reasoning in AI Agents?

Answer 11

Cognitive process where an AI agent analyzes information, applies logic, and makes decisions.

Reasoning is crucial for problem-solving and decision-making in AI systems.

Question 12

What is LangChain?

Answer 12

Open-source framework for building LLM-powered applications, including agents, with modular components.

LangChain provides tools to facilitate the development of AI applications.

Question 13

What is LangGraph?

Answer 13

Extension of LangChain for building stateful, multi-actor LLM applications with graph-based workflows, supporting cycles and HIL.

LangGraph enhances the capabilities of applications built with LangChain.

Question 14

Define Model Context Protocol (MCP).

Answer 14

Open standard for standardized interaction between AI models and external tools/data sources.

MCP aims to facilitate interoperability between different AI systems.

Question 15

What is the Agent-to-Agent (A2A) Communication Protocol?

Answer 15

Open standard for secure and interoperable communication between independent AI agents.

A2A protocols are essential for multi-agent systems to collaborate effectively.

Question 16

Define Retrieval Augmented Generation (RAG).

Answer 16

Technique to enhance LLM responses by retrieving external knowledge and providing it as context.

RAG helps improve the accuracy and relevance of AI-generated content.

Question 17

What is a Multi-Agent System (MAS)?

Answer 17

System composed of multiple interacting AI agents working together.

MAS can solve complex problems more efficiently than single-agent systems.

Question 18

True or False: Human-in-the-Loop (HITL) systems involve no human oversight.

Answer 18

False

HITL systems integrate human involvement for validation and input in AI workflows.

Question 19

What does Orchestration refer to in AI Workflows?

Answer 19

Coordinating and managing the execution of multiple components or agents in a complex workflow.

Effective orchestration is key to achieving desired outcomes in AI processes.

Question 20

What is a Hallucination in the context of LLMs?

Answer 20

When an LLM generates incorrect or nonsensical information.

Hallucinations can undermine the reliability of AI-generated outputs.

Question 21

Define Guardrails in AI.

Answer 21

Safety mechanisms or rules to guide AI behavior and prevent undesirable outcomes.

Guardrails are critical for ensuring the responsible use of AI technologies.

Question 22

What is Ontology in AI?

Answer 22

Formal representation of knowledge as concepts and their relationships within a domain.

Ontologies help structure information for better understanding and processing by AI systems.

Question 23

What company is behind LangChain and LangGraph?

Answer 23

LangChain, Inc.

This company provides tools for building LLM applications and agents.

Question 24

What does Google offer in the realm of AI agents?

Answer 24

Provides a platform and open-source ADK for building enterprise-grade AI agents with support for MCP and A2A, and a managed runtime (Agent Engine).

Google’s tools facilitate the development of scalable AI solutions.

Question 25

Which company provides AutoGen, Semantic Kernel, and Copilot Studio?

Answer 25

Microsoft ## Footnote These frameworks and platforms assist in creating multi-agent systems and integrating LLMs.

Question 26

What does Amazon offer for building AI applications?

Answer 26

Provides a managed service (Bedrock Agents) for building task-performing AI applications and an initiative (Nova Act) focused on autonomous web interaction. ## Footnote Amazon's services aim to enhance the efficiency of AI solutions.

Question 27

What is OpenAI known for in AI development?

Answer 27

Offers APIs and SDKs for building customizable AI agents with built-in tools, leveraging their advanced LLMs. ## Footnote OpenAI's offerings are widely used for developing AI applications.

Question 28

Who is behind the Claude family of LLMs?

Answer 28

Anthropic ## Footnote Anthropic focuses on creating safe and capable LLMs for reliable AI agent development.

Question 29

What framework did IBM develop for agent-based workflows?

Answer 29

BeeAI ## Footnote BeeAI is designed for scalable enterprise deployments involving AI agents.

Question 30

What is CrewAI?

Answer 30

Open-source framework for orchestrating role-playing, autonomous AI agents that collaborate as a 'crew.' ## Footnote CrewAI emphasizes teamwork among AI agents in various tasks.

Question 31

What is MetaGPT?

Answer 31

Open-source framework that simulates a software company using multiple AI agents in specialized roles following SOPs. ## Footnote MetaGPT showcases the potential of AI in business simulations.

Question 32

What is a strategy for enhancing LLM responses?

Answer 32

Retrieval Augmented Generation (RAG) ## Footnote RAG is used to improve accuracy and reduce hallucinations in AI outputs.

Question 33

What does Human-in-the-Loop (HIL) involve?

Answer 33

Involve human oversight and intervention at critical points in agentic workflows, especially for sensitive tasks. ## Footnote HIL is crucial for ensuring the quality and safety of AI outputs.

Question 34

What is the strategy of Multi-Agent Systems (MAS)?

Answer 34

Utilize multiple specialized AI agents that collaborate to solve complex problems. ## Footnote MAS leverages the strengths of various agents to address challenges more effectively.

Question 35

Define Phased Deployment.

Answer 35

Implement AI agents gradually, starting with pilot projects and scaling up based on performance and feedback. ## Footnote Phased deployment helps mitigate risks associated with AI implementation.

Question 36

What does Continuous Monitoring and Maintenance entail?

Answer 36

Regularly track agent behavior, performance, and resource usage, and update based on new data and requirements. ## Footnote Ongoing maintenance is essential for optimal AI agent performance.