Agentic AI Bootcamp: AI Agents with Python, n8n, MCP & RAG

Install essential tools like Python (pip, uv), Node.js, and a front-end stack (React, TypeScript, Vite, Tailwind), then learn Cursor, and explore Git and GitHub for version control.

Install python, pip, and the uv package manager, with optional pyenv for managing multiple versions. Learn to verify installations, compare pip and uv, and adapt to Windows, macOS, Linux environments.

Use LLMs to speed up installations by asking for the fastest pip installation commands, verify Python and pip setups across Windows, Mac, or Linux, and leverage ChatGPT to save time.

Install Node.js locally to run JavaScript outside the browser and power React with Vite and Tailwind. Explore npm and nvm for managing node versions on Windows, macOS, and Linux.

Learn the difference between git and GitHub, how git provides local version control and branching, and how GitHub hosts remote repositories for collaborative code projects.

Master ai fundamentals, including the differences between automation, ai automation, and ai agents, how llms use tokens and APIs, and the basics of prompt engineering and function calling.

Explore automation, ai automation, and ai agents, defining triggers and actions, comparing workflows from webhooks and Google Sheets to multi-agent orchestration with llms, tools, and a ceo.

Learn how large language models work, from parameter and run files to training phases—pre-training, fine-tuning, and reinforcement learning—plus tokens, token limits, context window, and local versus API deployment.

Learn how an API, or application programming interface, enables software systems to communicate via HTTP requests and JSON, using standardized input and output, endpoints, and language-agnostic, scalable access.

Explore how function calling lets language models use tools and APIs, including web searches, code execution, databases, memory, and email integration, to expand capability.

Explore test-time compute, or thinking, in llms as models reason step by step to improve math, coding, and science tasks, while noting slower performance for creative writing and trade-offs.

Learn prompt engineering basics by distinguishing user prompts from system prompts, structuring prompts with clear context and examples, and deciding when to use or omit system prompts.

Discover how to find the best ai models using artificial analyses leaderboards, open weights, and licenses, across image, video, and text models you can run locally.

Explore automation, ai automation, and ai agents, and how llms reason with tokens to produce outputs. Discover how api calls, system prompts, and prompt engineering affect cost, speed, and capabilities.

Build AI agents with the OpenAI playground and agent builder in Python or TypeScript, exploring guardrails, state variables, tools, and WeChat jet Kit integration with Next.js chat.

Explore OpenAI's agent kit, a low-code platform with a visual canvas to build, deploy, and optimize AI agents, including guardrails and the Python or JavaScript SDK.

Set up an OpenAI account and billing, explore the playground, test models from GPT-5 to nano, and manage projects, API keys, and pricing for agent kit development.

Build a mini AI agent using state variables, system prompts, and internet tools to enable named interactions and guarded responses.

Connect your agent to 8000 apps with the Model Context protocol as an API wrapper via MCP servers like Zapier, enabling Gmail and calendar workflows and date-time control.

Learn to build a RAC agent by uploading text files, creating a vector database, and using context-aware embeddings with similarity search to answer questions.

Create a more advanced workflow and conduct deep research for yourself, embracing the challenge as a valuable learning experience; you can retry this lecture at the course end if needed.

Build a deep research agent that generates keyword phrases, loops through them with a while node, and delivers a comprehensive, source-backed analysis plus a ten year valuation metrics comparison.

Next steps integrate a chat widget into ai agents and build the front end. Then add a pop-up chat bubble to a web page, from scratch in a coding editor.

Build and test chat widgets with the widget builder, browse the gallery, and upload a weather widget to publish it in production. Be aware of occasional widget bugs.

Learn to build a production-ready chat UI with chat kit in Cursor as a Next.js app, configure guardrails and moderation, connect a vector store, and deploy via chat kit workflow.

Build a Next.js web page with a pop-up ai chat bubble using cursor and chat kit, from scratch to localhost, with a comfort assistant.

Monitor agent costs by reviewing spend to date, token usage, and 87 requests, with spend around $0.35 and GPT five as the largest cost, suggesting GPT five mini for savings.

Explore templates, workflows, drafts, and saves to accelerate AI process creation. See how data enrichment templates perform web research, summarize findings, and output data for company info and Q&A.

Build an MVP chatbot using Python, an AI agent framework, and Pydantic AI with an OpenAI model, and expose it with a Gradio frontend.

Explore the tech stack for a deep research agent using Cursor, Python and pip, powered by Pedantic AI and Gradio, with multi-agent orchestration for real-time web research and markdown outputs.

Build a simple chatbot prototype with Pydantic AI and Gradio, using GPT five mini, docs-driven setup, and iterative debugging to create a stand-alone app.

Feed cursor the LMS folder context and tag documentation for every project by adding docs from a full URL, with the prefix and entry point, using pydantic.

Build a deep research agent in Python with a Gradio frontend that orchestrates multi-agent web research using DuckDuckGo, producing structured executive reports with sources and insights.

Learn how a multi-agent ai project uses a dot env secure key store, classification, planning, parallel deep dives, and a writer to generate structured markdown reports via gradio.

Publish your Python project on GitHub by creating a repository, uploading files, and committing changes, then share a Gradio app via a public URL while noting key security cautions.

Build local AI agents with open source models, connect to a local Olama server, and use LangChain, Flowise, and dockerized Postgres vectors for private retrieval.

Understand hardware essentials for running local AI models, including LMS and diffusion models, with emphasis on GPU VRAM, RAM, storage, CPU, and cooling.

Explore how Apple hardware enables local AI with unified memory for efficient llms and diffusion. Compare CUDA absence, upgrade limits, and model sizes from M1/M2 to M3/M4 Pro.

Install Ollama, download and pick suitable models, and understand vram and fp16/q4 options. Run a local server and call endpoints from the terminal to build ai apps.

Explore how LangGraph and LangChain power easy AI agent workflows with Flowise, showing background code execution, local runs, and tools for observation, evaluation, and deployment.

install Flowise locally with Node.js, explore its open source self-hosting interface, and use local or cloud models to build RAC agents and vector databases with Postgres.

Build your first mini workflow in Flowise, connecting a start node to direct replies, condition logic, and iteration to test the agent builder interface with local models.

Build a local rag agent with Flowise, a Postgres vector store, and llama, enabling memory, multiple knowledge bases, tool integration, and human-in-the-loop validation.

Prepare your data for a RAC vector database by converting to markdown or qa formats, removing unnecessary data, and optimizing chunk size and overlap for reliable retrieval.

Self-host AI agents on your machine, remaining uncensored and independent from big tech. Explore the long chain ecosystem with Docker, Postgres, and vector databases to build local applications.

Learn to deploy AI agents via OpenAI models and API brokers, self-hosted and online, with sentiment analysis workflows, rack apps, web scraping, and flow wise front ends for client projects.

Build an AI-powered sentiment analysis flow for customer support, capturing mood with a form input and routing via a condition node to tailored LLM responses.

Build a rag chatflow by web scraping, embeddings, and an in-memory vector store, using Brave Search API and HTML to markdown to answer questions from documents.

Export and import Jet Flow workflows as JSON, download and share them, and load saved flows in a new canvas to quickly replicate or distribute AI agent workflows.

Create a standalone RAC chatbot and embed it in web and apps, with Python, JavaScript, or curl, plus hosting, public access, and customization options such as title settings and colors.

Learn to build a Flowise tool-agent with a dual agent that connects APIs and LLMs in one workflow, using tools and memory.

Learn to build a dual agent with Pinecone as a hosted vector database, embedding and upserting document chunks, and enabling retrieval via a retriever tool.

Explore how system prompts and additional parameters shape AI agents, including memory, tool usage, and max iterations, to orchestrate tool calls like Brave search, calendar, and vector stores.

Host flow wise chatbots on render, from fork to web service, and learn pricing, environment variables, and persistent disk for client-ready deployments.

This lecture shows how to build a client-ready RAG chatbot with Flowise tool agent, covering document stores, embeddings, pinecone vector store, and a German medical assistant system prompt.

Refine the LangGraph chatbot on Replit by improving the German system prompt and training data. Embed a branded flow wise dock on a simple HTML page.

Embed a rag chatbot into a WordPress site using a vip code plugin, add an html snippet with the chatbot script, and place it in the footer for all devices.

Configure and monitor your LangChain app in Flowise with view messages, feedback, leads, rate limits, allowed domains, starter prompts, and Lang Smith integration.

Explore the marketplace templates to extend workflows. Validate queries with GPT-4 omni or a mini model, search a vector database, and loop back via the retriever with human-in-the-loop checks.

Explore a practical ai agent workflow using sentiment analysis to route to OpenAI or other models, and implement hosting, branding, and vector database integration with pinecone.

Learn to build ai agents with python and n8n, using local or api models, connect databases via docker, craft a rac agent with metadata, and create an email assistant.

Install n8n locally with Node.js and npm, learn global installation, then explore the interface, templates, and pre-built agents to build AI-enabled workflows, with self-hosting vs cloud hosting options.

Learn the basics of n8n workflows: define triggers and actions, explore diverse triggers (chat, form, mail, webhooks, MCP server), and wire them to AI agents and tools.

Install and run Supabase in Docker to use vector databases, embeddings, and SQL tables locally, then access a local instance on port 8000.

Create a local RAG workflow with an ai agent, llama and ollama, storing data in Supabase vector store using alarm embeddings, and querying with metadata-driven vector search.

Explore exporting and importing JSON workflows for sharing and reuse in n8n, including downloading JSON, viewing fields like name and node positions, and importing into a workflow.

Build an email automation agent that queries your database, retrieves contacts from Supabase SQL, and sends targeted emails via Gmail, using OAuth credentials and SQL integration.

Create a system prompt by defining a role, goal, and MCP tools. Include rules, style, output format, and variables, and test with ChatGPT using an optimizer.

Self-host n8n workflows and leverage telegram and webhooks as triggers, while comparing hosting options from Hostinger, render, and other cloud services for scalable automation.

Recap of n8n basics explains installing locally, configuring triggers and actions, and using AI models with HTTP requests, alongside hosting workflows with Docker and Supabase.

Master building HTTP requests, using get, post, and more; run JavaScript or Python code in a code node, scrape data, and build a telegram agent with webhooks and guardrails.

Learn how to connect any API using HTTP request nodes, get or post methods, and import curl to set up authenticated requests with headers and tokens.

Run custom code in workflows with the code node (JavaScript or Python, beta) and merge outputs using the merge node to enable complex AI agent workflows.

Scrape web pages with http request nodes, using robots.txt and sitemap checks, convert HTML to markdown, and store in a Pinecone vector database with OpenAI embeddings for a rag chatbot.

Build a Telegram AI assistant that triggers from Telegram, uses whisper for audio, and routes to an OpenAI agent with memory, calendar, and Gmail tools.

Design a CEO orchestrator that calls sub workflows and tools like Gmail, calendar, and research agents to build a large workflow.

Develop a talkative Telegram agent by cloning and expanding sub workflows, integrating Gmail and calendar tools, a web search agent, and automated social posts to orchestrate complex automation end-to-end.

Secure Telegram-based n8n workflows by restricting access with chat IDs and usernames, using if nodes to prevent unauthorized access to mails, calendars, and contacts.

Learn to build multi-agent systems with an orchestrator as a CEO that triggers sub workflows and sub agents on one canvas, including HTTP interactions.

Explore a multi-llm workflow that distributes a chat query to four language models, aggregates responses, and uses Gemini judge to synthesize the best insights into one comprehensive answer.

Build a chatbot frontend with cursor using webhooks, trigger workflows via http post, and receive ai agent responses through respond to webhook, optionally connecting Pinecone and OpenAI embeddings.

Explore essential n8n data manipulation techniques, including set note, edit fields, split out arrays, aggregate and merge items, and routing with if and switch nodes, plus practical database connections.

Show how to loop over items with the loop node to process and enrich data step by step, using a code node and AI agent to generate captions for marketing.

Explore n8n guardrails in react native workflows, using an AI node to check text for violations and sanitize personal data, secrets, and URLs with custom regex.

Learn to build an error monitoring workflow in n8n that emails alerts when a workflow fails, using the error trigger node for reliable debugging.

Publish a rack application as a standalone app with a public url by hosting the chat via an addon and enabling a chat trigger and authentication options.

Learn to integrate Nadan chatbots into websites using HTML, WordPress, and custom CSS branding, including embedding, production webhook usage, and branding customization on Replit.

Explore building flexible AI agents with Python and n8n by mastering HTTP requests, API integration, code nodes, Telegram bots, Pinecone, webhooks, and guardrails.

Explore the MCP model context protocol, its stdio and http transports, and how to connect MCP servers to cloud desktop for hosting, multi‑server setups, and ChatGPT with image generation.

Explore how the model context protocol equips llms with tools, resources, and prompt templates. Understand why MCP servers standardize and simplify API calls, enabling dynamic self-discovery and seamless tool integration.

Start with stdio for local development, offering fast, direct process communication, then switch to streamable http for production over http post with an optional server send event.

Connect a file system MCP server to Claude Desktop by editing the Cloud Desktop Config.json. Restart and test read, write, edit files and directories.

Learn to install multiple MCP servers quickly using the MCP installer, integrating various tools via cloud desktop config, and resolve setup issues with logs and config edits.

Discover how to quickly find MCP servers and clients on GitHub, explore glamor MCP servers and awesome MCP clients, and filter by language, framework, and stars.

Learn to connect MCP servers to Cursor as a client, using Zapier for free, and assemble tools like Gmail, Google Sheets, and Google Docs to automate workflows across hosts.

Build an MCP server in n8n and connect tools like Google Sheets to append and get rows. Trigger the server from any client while addressing local versus cloud hosting security.

Connect your MCP server to multiple hosts, Claude, Cursor, n8n, and Windsurf, using native MN, Cloud Desktop, and Google Sheets tools to trigger workflows and manage leads.

Learn to secure your MCP server with authentication by configuring bare or header authentication, creating credentials, and updating the config file to control access.

Learn to integrate your preferred tools, including call and workflow tools, code tools, and http requests, to build your MCP server with options like Supabase, Airtable, or AWS.

Discover how to connect ChatGPT with an MCP server using the model context protocol, via apps and connectors, including GitHub, Google Drive, and Zapier, with advanced settings and developer mode.

Build an MCP server to generate images by orchestrating workflows and HTTP calls, using flux or OpenAI models, converting base64 to files, and uploading results to Google Drive.

Activate a self-improving agent to persist context across cloud desktop, cursor windsurf, and addons by saving memories in a vector database and recalling them for cross-host conversations and meetings.

Install the n8n MCP community node to list and execute tools, and connect native and community MCP workflows in self-hosted or local environments.

Enable instance-level MCP access in an addon, connect a single instance, and manage selected workflows via webhooks or chat triggers with Google Sheets product data and cloud desktop integration.

Discover how Blender connects to Cloud AI via the Model Context protocol, enabling prompt-driven 3D modeling, scene creation, and direct Blender control.

Explore a GitHub repository overview of the Python MCP you built, including a calculator server, prompt templates, resources, a webhook integration, and a ready-to-clone setup with readme and cloud config.

Explore how to include resources and prompt templates in your MCP server, enabling customer playbooks and webinar-to-blog prompts with dynamic transcript variables.

Extend your server by adding more tools and workflows via webhooks, defining tools with descriptions, and wiring prompts, so you can trigger APIs, weather data, or mail actions.

Publish the MCP all-in-one server to GitHub, update the Readme, and upload calculator server files, prompts, docs, and config, while securing API keys in a dot env file.

Explore the model context protocol and MCP servers to connect tools, resources, and prompt templates across hosts like ChatGPT and cloud desktop, with secure authentication and flexible sdk options.