Sign in or create an account to watch the recording
Register once to get unlimited, free access to the free sessions
How AI tools actually work under the hood, the simple loop of text going in and text coming out that powers ChatGPT, Claude, Cursor, Copilot, and every AI product you use today.
What all the buzzwords really mean — MCP, tools, skills, agents, sub-agents, function calling, "agentic AI.
A mental model that ages well — so when the next AI feature launches next week, you can look at it for 30 seconds and know exactly what it is.
Every week there's a new AI buzzword — MCP, agents, sub-agents, skills, tool use, function calling, "agentic" this, "autonomous" that. Most explanations tackle them one at a time, use jargon on top of jargon, and leave you more confused than when you started.
This session does the opposite.
One mental model. Six hand-drawn diagrams. The entire AI stack, mapped — in plain English.
You'll walk through:
Why an LLM (the thing inside ChatGPT, Claude, etc.) is just a text-in / text-out box — and why that single insight unlocks everything else
How tools actually work: the AI literally writes instructions, real code runs them, and the answer comes back as text
What MCP really is once you strip the marketing — spoiler: a simple text protocol, nothing more
Why agents are just the text loop running a few times in a row
Why sub-agents are just AIs calling other AIs — and why it's turtles all the way down
No coding. No frameworks to install. No SDK shilling. Just a clear map of what's under the hood of every AI tool you use, so you can stop feeling behind every time something new drops.
Curiosity about AI — that's genuinely the only prerequisite
No coding background required. No programming, no math, no jargon assumed
A laptop or tablet for note-taking (optional — the session is diagram-driven and slides will be shared after)
No prior knowledge of MCP, agents, or LLMs required — we start from absolute zero and build the full picture in an hour
Amr is a software engineer at AWS, based in Berlin. 8+ years across frontend and full-stack, these days mostly shipping production systems that have to stand up to real teams and real traffic, not just look good in isolation.
He is founding engineer behind zomra.io, an online learning platform built end-to-end with AI-assisted workflows.
He is also the founder of stackmy.ai and tadween.me. Across all of these, AI isn't a side tool for him, it's the core of how he builds. He designs systems where AI handles execution under structured guardrails, and has shipped production software this way at scale. This course is a distillation of how he actually works every day.