Hello Reader,

Agentic AI is the new buzzword. Every YouTube video, LinkedIn article, and blog is about Agentic AI. In today's edition, we will go over what is Agentic AI, and why is it becoming so popular. To understand this, we have to see the evolution of LLMs, let's find out:

In the beginning, there was a single LLM, and we were doing prompting. This was significant first step, but it had the following challenges:

1. Static, pre-trained information
2. Not able to integrate project-specific data

Then came RAG to overcome these challenges. RAG became very popular because you can integrate your company data using embeddings. But RAG has the following challenges:

1. Once RAG gives you an answer, there is no way to validate, iterate, and refine the answer
2. Non-dynamic access to data, heavily biased towards embedded vector database
3. Not able to create dynamic flow to take actions

To overcome all the above challenges and then some, came Agentic AI. With Agentic AI, LLM agents can iterate, refine, and take actions (see the diagram with numbers).

The advantages of Agentic AI are:

1. Because you can invoke code, you can literally do anything from the LLM. You can even have multiple functions, and LLM can dynamically invoke them based on prompt
2. Iterative flow - LLM agent validates result and iterates and refine
3. Given the flexibility of calling an API or code, you can perform actions, e.g., book a vacation based on budget, send emails based on meeting recording, etc.
4. This is the fundamental flow; you can even invoke multiple Agents and multiple LLMs using Agentic AI, making it more powerful

Agentic AI on Whiteboard

I sat down with AWS director, Roland Barcia. He covered Big Tech interview tips, Gen AI, and we also did a whiteboarding session on Agentic AI with actual use case. Check it out, timestamps included for viewing conveience:

​

​

Salesforce CEO Marc Benioff has emphasized that the current generation of CEOs will likely be the last to manage an exclusively human workforce, predicting a future where humans and AI agents work together. He described AI agents as "digital labor," highlighting their role in significantly boosting productivity.

I am personally learning different agentic patterns, tools, and underlying infrastructure. In the coming weeks, I will share what I am seeing with customers in the trenches and some of the patterns I am finding helpful.

If you have found this newsletter helpful, and want to support me 🙏:

Checkout my bestselling courses on AWS, System Design, Kubernetes, DevOps, and more: Max discounted links​

AWS SA Bootcamp with Live Classes, Mock Interviews, Hands-On, Resume Improvement and more: https://www.sabootcamp.com/​

Keep learning and keep rocking 🚀,

Raj