|
Agentic AI Full Course with LangGraph: Build AI Agents & Automate Workflows (Zero to Hero) - Part 1
DSwithBappy
This is Part 1 of a zero-to-hero course on Agentic AI using LangGraph, taught by an instructor named Bappy. The video primarily outlines the full course plan, which is divided into seven phases covering: an introduction to agentic AI concepts, asynchronous programming and Pydantic, LangChain fundamentals, LangGraph components and workflows, memory/planning/monitoring, deployment and production engineering, and finally three end-to-end real-world projects (an agentic chatbot, a custom ChatGPT agent, and a multi-agent travel planner called TripMate AI). Prerequisites include advanced Python programming, basic generative AI/LLM knowledge, and tools such as VS Code, Git, Docker Desktop, and Postman.
|
|
This Kittl Agentic Feature Changes Everything
Philip Anders
Kittl has launched a new "Agentic AI" feature (currently in beta) that allows users to generate multiple t-shirt designs, layout variations, and Etsy mockup images from a single simple text prompt, with no complex settings required. The tool can take a reference image and create several unique design variations, adapt an existing design to different niches or sports, and even produce product mockup images specifying particular blank styles like Comfort Colors. Users are currently limited to generating a maximum of four designs at once, and the feature is also available to Printful users with a 50% discount on annual plans using the code "Agentic."
|
|
AI Agents Explained: Build Your First One in 2026
Roboverse
This video explains what AI agents are and walks viewers through building one using a free, open-source tool called Hermes from Nouse Research. Unlike chatbots such as ChatGPT, AI agents operate autonomously on a schedule, retain memory between sessions, and complete tasks without waiting for user prompts. The tutorial demonstrates deploying Hermes on a always-on cloud server (VPS) via Hostinger's one-click template, connecting it to AI models through Open Router, and linking it to a messaging app like Telegram so the agent can deliver things like daily news summaries and to-do lists directly to your phone. The presenter claims the entire setup takes about 10–15 minutes and requires no coding experience.
|
|
MCP vs Skills: Which Is Right for Your AI Agent and LLMs?
IBM Technology
The video explains two complementary methods for enhancing LLMs with additional context: MCP (Model Context Protocol) and Skills. MCP standardizes how AI models connect to external data sources and services by abstracting APIs into an LLM-ready format with managed authentication, making it ideal for real-time, permissioned data access (e.g., querying a CRM or checking infrastructure state). Skills, by contrast, are lightweight markdown-based packages containing reusable prompts, scripts, and metadata that teach a model *how* to perform a specific task in a consistent, repeatable way—addressing LLMs' non-deterministic nature when domain-specific formatting or workflows are required. The presenter concludes that both tools are open source, widely supported, and can be combined to provide complete context engineering for AI agents.
|
|
How to Start an AI Agent Business as a Teenager in 2026
Jake One Page
The video explains how teenagers can start an AI agent business in 2026 by building and selling automated services to local businesses using a no-code platform called Base44 Super Agents. The creator argues that older approaches like freelance gigs and chatbot services are becoming obsolete, and that the real opportunity lies in deploying AI agents that work continuously across multiple clients without requiring coding skills. The tutorial walks viewers step-by-step through setting up a Base44 workspace, configuring an agent's memory and tool integrations (such as Gmail, Notion, and HubSpot), and building a lead-response agent that automatically replies to inquiries within 10 minutes and sends follow-up emails. The video also promotes a free masterclass (normally $499) linked in the description for those wanting to learn more about building AI-powered businesses.
|
|
AI Agents For Beginners – OpenClaw Case Study
freeCodeCamp.org
This beginner-friendly course, led by Bumshad Manhatt (founder of CodeCloud), teaches AI agent development from the ground up—starting with LLM fundamentals like how transformers, tokenization, temperature, and prompting work, then progressing through AI system architecture and agent implementation. Students build four hands-on agents named Zippy, Savvy, Meshy, and Cody, culminating in a multi-agent system where Zippy orchestrates a research agent, memory agent, and coding agent. The course concludes with an in-depth case study of OpenClaw, a popular open-source AI agent, examining its architecture, agent loop, memory system, testing strategies, and security considerations. All API keys and sandbox environments are provided so learners can practice without infrastructure setup or unexpected charges.
|
|
Agents are slower than LLMs?
Caleb Writes Code
Agents are slower than LLMs because they rely on external tool calls—such as web fetches, API requests, database queries, or file transfers—that can take seconds to minutes to complete, whereas the LLM itself generates token requests in a fraction of that time. This creates an imbalance where expensive GPUs sit idle while waiting for tool execution results, wasting costly compute resources. To address this, the video explains the concept of disaggregated inference and KV cache offloading, where the key-value cache is stored externally during tool execution so GPUs can be freed up, with specialized hardware handling compute-heavy prefilling and bandwidth-heavy token generation separately. The video also notes that agentic workloads ripple down through every infrastructure layer—affecting chip utilization, storage, networking, and even data center energy budgets—though these optimizations are primarily relevant at cloud scale rather than for single-user local setups.
|
|
Give Your AI Agent a Second Brain (Gbrain + Hermes Agent)
Tonbi's AI Garage
The video introduces Gbrain, an open-source tool (25,000+ GitHub stars) created by Y Combinator president Garry Tan, which acts as a "second brain" for AI agents like Hermes by ingesting external knowledge—such as meeting notes, Obsidian vaults, and markdown files—that never passed through the agent's chat. The host distinguishes Gbrain from Hermes's built-in memory (which only recalls past conversations) and LM wikis (which cover specific domains), explaining that Gbrain stores knowledge about "your world" using a hybrid vector/keyword/knowledge-graph search over a local PGlite database synced from a Git-backed markdown repo, exposing 30+ tools via MCP. The video then walks through installation using Bun and an embedding API key (e.g., via OpenRouter), demonstrates seeding Gbrain with demo data, and shows Hermes successfully querying that external data—such as retrieving meeting preparation details from notes the agent had never directly seen.
|
|
Moms Are Making $250k/Year: How To Build An Audience That Buys Using The NEW AI Agent (Step By Step)
The Calum Johnson Show
The video features an interview with John Wu, founder of the creator monetization platform Stan, who explains how everyday people — including a single mother who made $250,000 in her first year and an 18-year-old who earned $100,000 in two months — are building profitable online businesses by consistently posting content and selling digital products. Wu argues that the single most consistent trait among top earners on Stan's platform is simply posting content every day, and that AI tools now dramatically reduce the workload required to do so. He describes Stan's AI agent as a way to compress the effort of content creation, making it feasible for anyone to build an audience and monetize it through guides, courses, or other digital products. Beyond the business case, Wu emphasizes that entrepreneurship through content creation builds personal agency and self-confidence, which he views as equally — if not more — important than the financial outcomes.
|