Agentic AI - Series 1

-

 

What is Agentic AI? 

Agents are nothing but a software programs which can think and act autonomously. This is often called "ReAct" model. 

Re - Reasoning.

Act - Acting based on that what they thought. 

Reasoning is not as same as Thinking. Because, reasoning is a logical process of deriving a conclusion based on the ideas.

E.g.:

Let's say you want to drive to your friends place. Based on the previous experience you might think it takes 30 minutes to reach the friends place - This is called "Thinking".

Now, the same route to your friend is analyzed by some GPS applications like Google/Apple Map and it does logical analysis by evaluating the current traffic and other attributes and shows the best possible route which might show the ETA as 15 to 20 mins - This is called "Reasoning".

How does the Agent works?

Agents are build on top of the LLM (Large Language Models). Agents can use various tools like web scraping, web search and other activities to complete the work assigned to it.


Imagine a task given to a agent.

  1. Task is "Prepare a coffee with less sugar".
  2. Agent understand the task (Reasoning).
  3. Next, it plans on how to make coffee and the ingredients and tools involved.
  4. It uses the tools available to the agent like milk, coffee, sugar and vending machine.
  5. Finally, it serves the coffee as requested by the user.
  6. User tastes the coffee and shares his feedback stating "Sugar is bit high".
  7. Feedback is accepted by the agent and improves the planning phase on how much sugar to be added.
  8. This is called "Reinforcement" learning - A continuous feedback loop for learning and improving the agent capability.  

Core components of an Agent:

  • Role of an Agent.
  • Backstory of an Agent.
  • Goal of an Agent.
  • Tools for the Agent.
Role - As it says what is the role played by the agent [or] who is he?
  • Stock advisor.
  • Customer support representative.
  • Health card advisor.
  • Loan officer. 
  • System administrator.
Backstory - Reveals the knowledge level of an agent. 
  • You are a experienced stock advisor who have 15+ years of experience in stock broking and funds. Through out the experience you have successful records of maintaining customer equity portfolios and stock recommendations based on the customer expectation. 
Goal - Task assigned to an agent.
  • Stock advisor agent has been assigned a task to suggest 10 stocks for a beginner who is new to stock market with capital investment of 1 lakh who needs a decent ROI of 5% per year.
Tools - Utilities used by the agent to achieve the task.
  • Agent starts exploring the investment tools and financial data for which it has access to to find the stocks based on the user expectation.
  • Tools can be internal or external based on the organization.
  • Can be a simple web scraping of Yahoo finance data to individual companies PnL, upcoming projects to suggest better stocks.

    How does agent performs the task?

    This is core of agents. Agent highly rely of LLM (Large Language Model).

    A Large Language Model (LLM) is a type of AI model trained on massive text datasets to:

    • understand language
    • generate text
    • reason
    • answer questions
    • write code
    • use tools

    Examples of LLMs include:

    • GPT‑4, GPT‑4o (from OpenAI)
    • Claude (Anthropic)
    • Gemini (Google)
    • Llama 3 (Meta)
    • Mistral (Mistral AI)

    Comments

    Post a Comment

    Popular posts from this blog

    K8s - ETCD

    -
    Image
      etcd is a "strongly consistent , distributed key-value store". Why etcd? 1. Consistency : Since the API server is the central coordination point of the entire cluster; strong consistency is essential. It would be a disaster if, say, two nodes tried to attach the same persistent volume over iSCSI because the API server told them both that it was available. 2. Availability: API downtime means that the entire Kubernetes control plane comes to a halt, which is undesirable for production clusters. The  CAP theorem  says that 100% availability is impossible with strong consistency, but minimizing downtime is still a critical goal. 3. Consistent Performance: The API server for a busy Kubernetes cluster receives a fair amount of read and write traffic. The secret behind etcd's balance of strong consistency and high availability is the  Raft algorithm . Raft solves a particular problem: how can multiple independent processes decide on a single value for somethin...
    Read more

    Agentic AI - Series 3

    -
    Image
                                                                          In this blog we will see how to build a simple agent without using any LLM model. Simple flow of an agent: Agent takes input -> Based on the input, it decides which tool to use -> Performs operation and send the output back to the user. An AI agent framework is  a set of tools, libraries, and structures that simplifies building, deploying, and managing autonomous AI agents. But, here we are going to build an agent without using any framework and LLM's. Agent functionality is to perform "Addition" and "Subtraction". Agent uses 2 python functions (Tools) to perform addition and subtraction operations. Finally, it send the output to the user. We are going to implement the above discussed functionality vi...
    Read more

    Agentic AI - Series 4

    -
    Image
                                                                       In this blog, will how to create a simple agent using AI framework like Langchain with a simple LLM model  gpt-4o-mini. from langchain_openai import ChatOpenAI from langchain_core . tools import tool import os # Creating a tool @ tool def add_numbers ( x : int , y : int ) -> int :     "Add two numbers"     return x + y @tool is a decorator to define a tool. This is a simple tool to add 2 numbers. The below line creates a Langchain LLM wrapper around on the OpenAI chat mode 'gpt-4o-mini' # Binding the tool to a model llm = ChatOpenAI( model = "gpt-4o-mini" ).bind_tools([add_numbers]) response = llm .invoke( "What is 5 + 7?" ) When we print response, is the expected outcome? Is it 12. NO...
    Read more