What is Agentic AI? Important For GEN AI In 2025

Summary

In this video, Krish Naak explores the difference between Generative AI and Agentic AI, introducing the concept of autonomous AI agents that perform complex workflows to achieve specific goals. He demonstrates how tools like LangChain, F-Data, and LangFlow can be used to build agentic AI systems for real-world applications, such as stock market analysis. Through a practical example, he shows how these agents interact with external APIs and make data-driven decisions. The video highlights the future potential of Agentic AI in automating business processes and decision-making.

Takeaways

• 😀 **Generative AI vs. Agentic AI**: Generative AI focuses on creating content based on user queries, while agentic AI works autonomously to achieve specific goals with complex workflows.
• 😀 **Autonomous AI Systems**: Agentic AI systems are autonomous, meaning they can perform tasks independently without human intervention, unlike generative AI which mainly generates content.
• 😀 **Complex Workflows in Agentic AI**: Agentic AI enables the creation of complex workflows that can integrate multiple AI agents and external tools to achieve business outcomes.
• 😀 **Frameworks for Agentic AI**: Tools like LangChain, LangFlow, FData, and Microsoft Autogen are pivotal in creating agentic AI applications, enabling the integration of external sources and APIs.
• 😀 **The Role of Tools in Agentic AI**: External tools like Google Search, Wikipedia, and financial data sources are essential in enriching the performance and outputs of agentic AI systems.
• 😀 **LangFlow and LangGraph**: LangFlow is a no-code tool for designing agentic AI workflows, while LangGraph helps in building complex agent-driven solutions by combining multiple agents and models.
• 😀 **FData Framework**: FData allows the creation of agentic AI systems for specific domains, such as finance, where it can integrate multiple tools and models to make data-driven decisions.
• 😀 **Practical Example - Stock Analysis Bot**: A financial agent is built using FData, which can automatically gather data on stocks (e.g., Tesla vs. Nvidia), analyze it, and provide a recommendation.
• 😀 **Learning Curve**: Developing agentic AI applications involves learning to integrate various tools, frameworks, and APIs to build intelligent, autonomous systems that can make decisions.
• 😀 **Future Potential**: Agentic AI is gaining popularity, and its ability to replace traditional software-as-a-service (SaaS) applications makes it a promising technology for future business and development use cases.

Q & A

What is the main difference between Generative AI and Agentic AI?

Generative AI focuses on creating content based on user queries, using LLM models to generate text, images, or other forms of media. Agentic AI, on the other hand, refers to autonomous AI systems that perform tasks independently to achieve specific goals, integrating multiple tools and working together in complex workflows.

What does Generative AI aim to do?

Generative AI is designed to create content based on given prompts. It uses LLM models to generate responses, such as writing articles, poems, or answering queries, with the goal of content creation rather than solving a specific task.

How does Agentic AI function differently from Generative AI in terms of workflows?

Agentic AI works by executing complex workflows autonomously. It can integrate various external tools and systems to achieve a specific goal, and it can improve its performance over time without human intervention. In contrast, Generative AI focuses solely on content generation and does not involve executing complex workflows or tasks.

What are some common frameworks used to build Agentic AI applications?

Some popular frameworks for building Agentic AI include LangChain, F-Data, Microsoft Autogen, and LangFlow. These frameworks allow developers to create autonomous AI agents and design workflows that integrate multiple tools and external systems.

What is the role of LangChain in Agentic AI?

LangChain is a framework that supports the integration of external tools and APIs with LLMs. It allows developers to build agentic AI applications that can query external sources like search engines or news APIs to gather real-time data and make decisions based on that data.

What does F-Data offer for Agentic AI development?

F-Data is an open-source framework that enables the creation of different AI agents, such as financial analysts, legal agents, and marketing agents. It allows integration with various LLMs and external APIs, and supports building autonomous systems that interact with these tools to achieve specific tasks.

How does LangFlow make building Agentic AI easier?

LangFlow is a no-code platform that simplifies the development of Agentic AI applications. It allows users to design complex workflows through drag-and-drop tools and automatically generates code for integration with various APIs and LLM models.

Can you provide an example of an Agentic AI application?

An example of an Agentic AI application is a financial analysis agent that compares stocks, such as Tesla vs. Nvidia. The AI queries stock data, gathers recent news, and generates recommendations based on the current market conditions. Multiple autonomous agents work together to perform these tasks within a workflow.

What is the significance of integrating multiple tools in Agentic AI?

Integrating multiple tools in Agentic AI allows the system to access diverse external data sources and services. For example, an AI might use a finance API for stock prices, a news API for sentiment analysis, and a search engine tool to gather relevant real-time information, all working together to achieve a business goal.

Why is Agentic AI expected to replace traditional SaaS products in the future?

Agentic AI is expected to replace traditional SaaS products because it enables fully autonomous workflows that can perform tasks without human intervention. It can execute complex business processes and adapt over time to improve performance, providing more personalized and efficient solutions compared to static SaaS products.