Scribd
Upload
2 views22 pages

Module 4 Notes

The document discusses AI agents, focusing on their customization, capabilities, and examples of agentic systems. It outlines the differences between single-agent and multi-agent systems, emphasizing features like reflection, tool use, memory, and planning. Additionally, it highlights various AI agents and their applications in tasks such as essay writing and game playing.

Uploaded by

raman barnwal
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2 views22 pages

Module 4 Notes

The document discusses AI agents, focusing on their customization, capabilities, and examples of agentic systems. It outlines the differences between single-agent and multi-agent systems, emphasizing features like reflection, tool use, memory, and planning. Additionally, it highlights various AI agents and their applications in tasks such as essay writing and game playing.

Uploaded by

raman barnwal
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

AI Agents

Customize LLMs
Data LLM GenAI Applications
• Text
• Images
• Speech
• Structured Pre-training Model Adaptation
Data

Retrieval Agents with


Prompt Supervised
Augmented Tools and
Engineering Generation Workflow Fine-tuning

Improves
Shapes outputs knowledge
grounding
Agentic Systems

Autonomous entities that perceive their environment, reason about


actions, and perform tasks

Embedding models within


autonomous decision loops
Increasing System Capability Through
Architectural Framing
Approach Primary Capability Added
Prompt Engineering Behavioral steering
RAG Knowledge grounding
Agents Decision autonomy

Better Instructions
→ Better Information
→ Better Actions
Non-Agentic Reasoning (Zero-Shot) Agentic Reasoning

• Please type out an essay on topic X • Write an essay outline on topic X


from start to finish in one go, without • Do you need any web research?
using backspace.
• Write a first draft.
• Consider what parts need revision or
more research.
• Revise your draft.
Coding Benchmark (HumanEval)
Agentic Systems
Single-Agent Systems Multi-Agent Systems

One
Multiple
Autonomous
Interacting
Decision
Agents
Entity
Examples of AI Agents
AI Agents in Smallville
AI Agents That Can Play Minecraft
OpenAI’s New Robot
Figure 01

Video Source: [Link] 12


Other AI Agents
• AutoGPT: [Link]

• AgentGPT: [Link]

• BabyAGI: [Link]

• Godmode: [Link]
Single-Agent Systems
Single-Agent Systems

• Reflection

• Tool Use

• Memory

• Planning
Reflection
Reflection
Tool Use
Memory

Abstract
New Dialog
RAG

Image Source: [Link] Image Source: [Link]


Planning
• Request: “Please generate an image where a girl is reading a book, and her
pose is the same as the boy in the image [Link], then please describe
the new image with your voice.”

Example adapted from HuggingGPT paper


Memory
Planning
(Goal → Steps)

Memory Tool Use


(State/Context) (Act/Observe)

Reflection
(Evaluate/Revise)
Recap

• Agentic Systems Overview

• Single-Agent Systems

• Building Single-Agent Systems

You might also like