Lectures | LaunchLab

Two lectures

Each lecture runs for an hour and works on its own; if you book both, LL 101 comes first. No coding experience is needed. To build along on their own laptops, students need a paid Claude plan; without one, they follow the build on the main screen.

LL 101 Build and publish a website with AI

Length: 60 minutes
Taught: At ESSEC Business School, to more than a hundred master’s students
Learning outcome: A working website for each student who builds along, and the steps to put it online
Prerequisites: No coding experience. To build along, a paid Claude plan (Pro or higher) for Claude Design and Claude Code
Materials: Claude Design, Claude Code, tiiny.host, GitHub, Vercel, Supabase

Students design a site in Claude Design, have Claude Code build it, and put it online with tiiny.host. After that we show how a team keeps a site up to date with GitHub and Vercel, and how logins and user data work with Supabase.

Running order (approximate)
Time	Part	What happens
0:00	Introduction	Who we are and what students will build
0:05	Part 1: Design, build, deploy	Claude Design for the look, the handoff to Claude Code, then a first launch on tiiny.host
0:25	Part 2: GitHub and Vercel	Saving and sharing work on GitHub (commit, push, pull request, merge), then automatic deploys with Vercel
0:42	Part 3: Logins and data	How sign-up and log-in work, what a token is, and Supabase as a backend
0:52	Task and questions	A take-home build task and a short feedback form

A prompt students type at the end of Part 1

“Put all my website files into one folder and zip it, with index.html at the top level, so I can upload it to tiiny.host.”

From the current version of the slides. Select a slide to see it full size.

Slide 12. The Part 1 workflow.
Slide 15. Git terms in plain English.
Slide 27. How logging in works.

LL 102 Automate everyday work with AI agents

Length: 60 minutes
Status: Not taught yet. We’re looking for a first class to run it with.
Learning outcome: One scheduled task, set up and run once in class
Prerequisites: No coding experience. To build along, a paid Claude plan (Pro or higher) for scheduled tasks, and a Google or Notion account to connect
Materials: Claude, connectors (MCP, the standard that lets Claude use other apps), scheduled tasks, Notion

Students connect Claude to apps they already use, then turn a job they do every week into a scheduled task that runs by itself. The last part shows how several agents can split up a bigger job.

The worked example is Yibo’s own morning briefing. It reads Gmail, Google Calendar, WhatsApp, Google Drive and Toddle (a school coursework platform), then writes one page to Notion. For a marketing class, the same pattern covers a Monday round-up of competitor news or a Friday campaign report.

Planned running order
Time	Part	What happens
0:00	Introduction	Who we are, and the difference between chatting with AI and handing it a job
0:05	Part 1: Connect	Connectors and MCP, when an API is needed instead, and connecting a first app in class
0:20	Part 2: Schedule	Writing a prompt that runs without you, then saving it as a scheduled task and running it once
0:35	Part 3: Orchestrate	Multi-agent systems, the daily briefing, and what not to automate
0:52	Task and questions	Setting the at-home check of the next morning’s run, questions, and a short feedback form