Skip to main content

13. AI App Palooza

Note: The video covers material not in the guide below — please watch in full.

Action Step

Complete this before moving on.

Watch the full video. Then pick one thing — something from work or your personal life — and ask yourself: how could I get Claude Code to do this? Start a session, explore the idea, and see how far you get. If you hit a breaking point where the output isn't good enough, create a skill to teach it what good looks like. That's the loop.

Comment in Slack

Post your answer in your onboarding channel.

What was your biggest takeaway(s) from this training?

Training Guide

This is the closing training for the Claude Code course. No demo, no follow-along — just the mindset Yasin wants you to walk away with.

The Question to Keep Asking​

The single habit that separates someone who took an AI course from someone who actually uses AI: always be asking yourself, "How can I get Claude Code to do the work I'm working on?" Can I use an API? An MCP? A CLI? Whatever tools are available — the instinct should be to reach for them first.

Even if the model isn't smart enough to handle something perfectly today, give it three to six months. These models are on an exponential curve — they're writing code for themselves and getting better automatically. And even right now, if Claude can't do 100% of the work, it might get you 20%, 50%, 70% of the way there. That alone changes the math on what's worth attempting.

Beyond Work​

Yasin rattles off examples to stretch your thinking beyond LeanScale work:

  • Apple Health + Steps app — Claude recently got an integration with Apple Health. Build a custom dashboard for your steps, or combine family members' data into one view.
  • Unified daily brief — A skill that checks your to-do list, Notion, Teamwork, and calendar, then gives you one consolidated brief every morning.
  • Meeting prep agent — An agent that researches all attendees and pulls relevant docs 30 minutes before a meeting, then sends you a summary.
  • Health tracking agent — Tracks your Apple Health data weekly and flags trends like declining sleep or missed activity goals.

Yes, apps exist that do some of this. The point is that you have complete flexibility to build exactly what you want, tailored to exactly how you work.

Skills at the Breaking Points​

Most people approach AI as a one-shot thing — one prompt, one output, done. But real work is never one-shot. Even when you do something yourself, you're subconsciously asking questions and iterating through a workflow.

The skill here is identifying breaking points — the moments where Claude's output isn't quite right — and creating a skill file that teaches it what good looks like. Want your meeting briefs in a specific format? Build a skill with a reference document. Want your research compiled a certain way? Same thing.

Then the meta-move: make a skill for how to make great skills. Use Claude to do evidence-based research, launch explore agents and research agents, compile the findings, and build that into a reusable asset. Every skill you create is a compounding asset that makes everything after it faster.

The Meme​

Yasin closes with a meme: you meet someone named Jason, immediately forget his name because your brain is full, and then ask Jason if he wants to hear about how you're using Claude Code. That's the life now. Multiple people on the team — Yasin, Jake, Anthony — have lost sleep because once you realize you can build anything, there aren't enough hours in the day. Ten agents running in parallel, ideas faster than you can execute them.

The limits used to be your knowledge and your skillset. Now they're your creativity and your imagination.