For a deeper discussion on RAG and building AI Agents, let's talk here. RAG is evolving—simple retrieval isn't always enough anymore. In this video, I break down Agentic RAG, why reasoning before querying a database is crucial, and how it leads to more intelligent AI systems. We’ve heard terms like KAG, Graph RAG, and Agentic RAG, but what do they actually mean?

In this video, I break down the essential concepts of inserting data into a vector database, including embeddings, loading, metadata, and text splitting. These are critical for creating effective Retrieval-Augmented Generation (RAG) systems and building a strong knowledge base. Using tools like n8n and Pinecone, I’ll guide you step-by-step through practical examples to help you master these foundational skills. Whether you're new to vector databases or looking to improve your AI agent systems, this tutorial is for you! Let me know what you guys think!

Hey guys! Thanks for all of the support on my latest video where I built the ultimate assistant. @Lenny Peng asked me about how to create a draft as a reply, and I figured I'd address it as a post since many of you may be looking for this functionality. The solution is within the Create Draft tool, you will add an option called 'Thread ID'. The logic here will be the same as the Reply tool where the email agent must first get the message's so it can plug in the thread ID. IMPORTANT You should be able to just add this into the existing 'Create Draft' tool, but you will have to refine the prompting of course. As you can see in the 4th and 5th image, even if the thread ID is left blank (meaning we aren't replying to an email), it sill creates a draft. However, it did make the subject seem like this is a reply. Like I said, introducing this will add some prompting elements. You could even leave the 'Create Draft' tool with no Thread ID option, and create one 'Draft as a Reply' with the Thread ID option. Similar to the 'Create Event' and 'Create Event with Attendee'.

Want to dive deeper into this build and discuss how to enhance it? Join AIS+ In this video, I build the ultimate AI personal assistant using a team of AI agents! A centralized assistant manages four specialized sub-agents, each handling a specific area—making the system more efficient and scalable. This approach ensures that no single agent gets overwhelmed with tasks and tools, leading to better automation and smarter decision-making. All of this is built in n8n using no code! Watch to see how it works. IMPORTANT: Import the workflows by opening up a new workflow, click the three dots in the top right, and select "Import from file". You must make sure when you are setting up the workflows as a tool for the main agent that you are choosing the correct workflow within your n8n environment.