Understanding Agentic AI: Features, Impact, and Practical Implementation

Topics Covered:

1. Introduction to Agentic AI

   • Rapid advancements in AI technology
   • Difference between generative AI and agentic AI

2. Impact of Agentic AI on Jobs and Skills

   • Job market transformations
   • Required skills for agentic AI (Machine Learning, Deep Learning, NLP, etc.)

3. How Agentic AI Works: Concept and Examples

   • Explanation using real-life agents (IRCTC, LIC agents)
   • Role of AI agents in decision-making and execution

4. Key Features of Agentic AI

   • Autonomous decision-making
   • Goal-oriented execution
   • Feedback-based improvement
   • Adaptability in changing scenarios

5. Comparison: Generative AI vs. Agentic AI

   • Generative AI focuses on content creation
   • Agentic AI is goal-driven and executes actions

6. Technological Implementation of Agentic AI

   • Tools and libraries (MCTS, Ray, Lanchain, etc.)
   • Use of AI in different industries (finance, healthcare, customer support)

7. Practical Demonstration: Andrew Ng’s Vision Agent

   • Introduction to Vision Agent AI
   • Image and video analysis capabilities
   • Real-world use cases in e-commerce and automation

8. Future of Agentic AI and Career Opportunities

   • Growing demand in tech industries
   • Importance of upskilling in AI-related fields

9. Conclusion and Engagement

• Summary of Agentic AI’s importance
• Encouragement to learn AI skills for career growth
• Invitation for questions and discussions.

Hello everyone, welcome to the blog. In today's post, we will understand agentic AI. Till now, we all are very well aware that AI is developing very fast, and technological advancements are taking place.  

We have heard a lot about generative AI and have been using it, whether it is ChatGPT or Google's Gemini. But now, what is this new technology of agentic AI? What impact will it have on the job market? Will new job opportunities be created? What skills will we have to learn for this agentic AI? We will understand all these things in today's session. And not only theory—we will also explore its practical implementation.  

For this, we will look at the work of Andrew Ng, a well-known figure in artificial intelligence, data science, and machine learning. He has been the head and co-founder of Google Brain and is also the CEO of his own company, Lending AI.  

The agentic AI that he has developed will be demonstrated with a practical implementation in today's session. This discussion is crucial because as soon as agentic AI started emerging, the founders and CEOs of major companies began discussing it—whether it is Meta, Google, NVIDIA, or Microsoft.  

The R&D divisions of every major company prioritize discussions on technology. Mark Zuckerberg mentioned in a recent interview that the new acceleration mode in AI will focus on agentic AI. Google's Sundar Pichai stated that agentic AI will introduce a new productivity model in the job market that could reshape the economy.  

NVIDIA’s CEO said that agentic AI has begun its rise, possessing the power to automate tasks completely. Microsoft's Satya Nadella also introduced an AI agent that the company is developing, specifically for customer supply chain management.  

So, what exactly is agentic AI? Before diving into the term, let's break it down simply. Agentic AI is also referred to as AI agents.  

To understand the concept, let’s consider the word “agent.” In our daily lives, we often use agents. For example, IRCTC agents help us book train tickets. We tell them our destination and preferences, and they handle the search, booking, and finalizing of the ticket.  

Similarly, LIC agents assist with financial investments. We communicate our financial goals, and they find suitable policies, execute the tasks, and ensure our objectives are met. In simple terms, an agent is someone or something to whom we delegate a task, and they carry it out on our behalf.  

Now, applying this concept to AI, agentic AI functions as an autonomous agent that can plan, execute, and refine tasks based on goals. It has four key features:  

1. **Autonomy** – It can make decisions without human intervention.  

2. **Goal-Oriented** – It is designed to achieve specific objectives.  

3. **Feedback-Based Improvement** – It continuously learns and optimizes performance.  

4. **Adaptivity** – It can adjust to new environments and changing conditions.  

With these capabilities, agentic AI is distinct from generative AI. While generative AI, like ChatGPT, generates content based on inputs, agentic AI focuses on achieving a defined goal with structured planning, execution, and feedback loops.  

For example, if you want to invest in stocks, generative AI can provide comparisons and news. However, agentic AI would analyze real-time data, create an investment strategy, execute trades, and adjust based on market changes.  

Companies are integrating agentic AI into various applications—finance, healthcare, customer support, and more. This is why industry leaders are prioritizing research in this field.  

The rapid rise of agentic AI also creates new career opportunities. If you want to enter this field, skills such as machine learning, deep learning, Python, statistics, computer vision, and NLP will be essential.  

Semiconductor giants like NVIDIA, AMD, and Intel have already begun focusing on agentic AI, along with cloud providers like AWS, Google Cloud, and Snowflake. The application side of AI is now gaining traction, with companies developing AI-driven applications across industries.  

To understand this better, let’s explore Andrew Ng’s Vision Agent, an AI agent specifically designed for image and video analysis.  

Vision Agent can be used for tasks such as identifying damaged products in e-commerce returns, verifying product colors against orders, and even analyzing video content to extract relevant clips. For example, if you upload an image of a sports scene and ask Vision Agent how many people are in the picture and which sport they are playing, it will analyze the image and provide accurate results. Additionally, Vision Agent generates the corresponding code for the analysis, allowing developers to integrate it into their applications.  

This technology extends to videos as well. Suppose you have a 5-minute football match video and only want to extract the clip of Ronaldo’s goal. Vision Agent can analyze the video and generate that specific clip, making content analysis highly efficient.  

Agentic AI has the potential to transform industries by automating complex decision-making processes, reducing human intervention, and improving efficiency.  

The future of AI is shifting beyond content generation to autonomous decision-making systems that can act, learn, and adapt in real-time. Understanding and developing skills in this area will be crucial for staying ahead in the evolving job market.  

Let me know your thoughts and questions in the comments. Looking forward to our next discussion. Stay curious and keep learning.