// transcript — 67 segments
0:00 What is an AI Agent?
0:29 Hello student and uh who are interested in learn agentic AI.
0:41 So in this session we will see what is agent basic agent and which component is involved in
0:49 agent. So what is LLMbased agent? LLMbased agent is AI agent where
1:00 decision making brain is large language model means LLM instead of fixed rules.
1:13 We can create a agent without LLM also. So for without LLM we use fixed rules for decision making
1:22 and with LLM we are keep our brain means LLM for making a decision.
1:33 So how we can define our definition will be become like LLM based agent use AI model to
1:42 understand think and response like human. So LLM should be understand what user
1:54 input should be there it think and then give response like a human.
2:03 So which component is in hall in LLM agent? The environment should be there
2:12 means where agent has to be work. It may be like chatbot, window,
2:17 application, terminal. Then sensor means input. It may be text base. It may be some values.
2:27 Then llm memory it is optional sometime memory is not required but whenever we have to
2:35 store the history about chart we should have to be introduce memory then output.
2:44 So that are five component I have mentioned out of four is compulsory and
2:50 one is optional and the memory is optional. I have mentioned here if we
2:57 need a take a data or push a data into database or keep uh histoologgical
3:04 uh communication so that type memory should be there. So four components
3:10 should be needed environment sensor LLM and accumulator means output should be
3:15 Simple vs LLM-based Agents
3:18 there. So what the architecture? So architecture we we can see here.
3:27 So basic architecture about LM user input LLM thinking and understanding decision making and output
3:35 that is a basic architecture very basic agent I'm giving to demonstrate you
3:44 then uh a real life example. So explain AI agent give explanation
3:49 summarize the text it will give summarization about text ask opinion
3:57 give reasoning that is the we can say uh the LLM agent will be do there then
4:03 comparison about simple agent and LLM agent that should be needed uh
4:08 understanding about basic simple agent means agent without LLM rule
4:16 How the decision will be there? The decision with help of rules, conditions
4:23 and in LLM based agent decision will be taken by LLM. No fixed condition is
4:30 there. LLM will be understanding and thinking and give response. Learning the
4:36 simple agent no learning is there. Llm learning is possible.
4:41 Language understand in a simple agent. No. In LLM agent language can be
4:49 understood with here. Flexible low. Here flexible is high. Example calculator.
4:56 The calculator is also one agent. Here we have giving the values with fixed
5:02 values. It give the correct answer whatever the calculation about that
5:08 values. And in intelligent example we give the chart GPT. I mention here chart
5:14 GPT because it is very known to all of you. Cost is very low. Cost is high.
5:22 Intelligence is none in agentic uh simple agent or rulebased agent and
5:27 intelligent or LLM based agent intelligence system. So on one line com comparison
5:38 simple agent react using rules while intelligent agent think using AI model.
5:48 So AI agent it take input response and produce output and simple agent use rule
6:01 So we will take one example. Suppose this is one example of uh llm sorry uh agent which is a rule
6:11 base. If uh we define one function simple agent input is a text and uh
6:23 if hello in text in lower return hello student it will AI it return AI means
6:30 Building a Rule-Based Agent (Code)
6:31 artificial intelligent return I don't know. So we will try run this code
7:43 So what is in AI uh in calling? So that's why it is called as AI means
7:48 artificial intelligence. Suppose we change here input we can give
7:58 what is bi. So answer would be I don't I don't know because this is a rule base
8:15 it is not understanding what exactly word and all this thing now we can give
8:23 another one example which is It use local LLM that's why it takes
9:25 some time. AI or artificial intelligence report to simulate our human because this is give
9:33 long answer. You can comparison about answer of previous one and this one it
9:38 give long answer. Now we will test So it give the answer also same because
9:56 we here we use LLM it think about BI and give the answer about the question.
10:02 BI stand for business intelligence which refer to the process of collection
10:07 analyze presenting data to business make informed decision. So perfect. So this
10:13 is a basic comparison about rule LLM and LLM. So I will explain the code here.
10:19 Now uh import here for local system I can mention then uh design one function simple l
10:41 agent here uh response is chart model this is herum test model because this model has
10:45 Installing Ollama & Hermes3
10:52 to be installed locally in my system. In a system message I have given role
10:57 system content your simple teaching agent answer is in one shot role user content
11:07 return uh message and content and here I have to be call this agent. So this is
11:16 uh basic comparison and how to agent will be created. So baseline about this
11:21 session is simple agent follow instruction and intelligent agent
11:26 understand the instruction. So please be continue with us, share
11:33 this video, like and subscribe the channel and we will discuss
11:39 topics which has to be related to the