Yohan Lee

wanting to be a philomath

[Gregor Zunic @gregpr07]

Breaking: Browser Use + Jev = Ultrafast ⚡

Findings flights took 7s and cost only $0.0039 🤯

> new action space every step
> DOM state space
> small LLM fallback to type

(this video is at 1x speed btw)

Built a tiny open source browser agent. try it below ↓

[Yohan Lee @yohan0405]

Watched @gregisenberg’s WebMCP episode → explored an agent mystery shopper with Claude Code → built this game in Codex after his “make a game” idea (#9). Five travel scenarios. One confident AI. You decide when to trust it. Can you get 5/5? https://t.co/CMUOS0t7jO

[Yohan Lee @yohan0405]

@gregisenberg Tried #9! Made a tiny game where you supervise a very confident AI travel agent. Five fictional situations: approve it, stop it, or ask for proof. Your WebMCP episode sparked my agent mystery shopper project. This is its playful front door. https://t.co/CMUOS0t7jO

[Yohan Lee @yohan0405]

What if your kid's wish became a tiny constellation?

CLOUDWORKS II: choose a card, turn the crank, open a little sky.

Designed & animated with Astra + Blender. Original sound. 3D concept; physical prototyping still ahead.

What should the next wish become?

[Yohan Lee @yohan0405]

@gregisenberg Building a public WebMCP demo for a client or launch?

I’m offering a $149 pilot: one agreed task, 3 supervised browser attempts, evidence + recommendations. Payment after delivery.

Reply with a URL and the task that must work. We’ll confirm scope first.

[Yohan Lee @yohan0405]

A @gregisenberg WebMCP episode inspired me to build an agent mystery shopper.

One lesson from testing my own scanner: a tool can return success and still give the wrong result.

So I’m checking task outcomes against explicit requirements. #WebMCP

[Yohan Lee @yohan0405]

@gregisenberg A synthetic check exposed the problem: my scorer gave the same 99/A to a correct $684 total and an incorrect $684,000 total. That score did not validate the outcome.

I rebuilt the test around explicit task requirements. Sample: https://t.co/YsDC7s9G5X

[Yohan Lee @yohan0405]

Family Stay Compiler links:

Live: https://t.co/bJCqbw6H61
Code (MIT): https://t.co/S8C0yDGz4E
Our report: https://t.co/9DwkLQbt6C

87/A is provisional: QA found scoring issues. It does not verify end-to-end task completion.

[Yohan Lee @yohan0405]

2 adults. 3 kids. "No rooms available."

What if the search box is the problem?

I built Family Stay Compiler for OpenAI's WebMCP Challenge: keep the whole family, find a multi-room plan.

@OpenAIDevs #WebMCP
https://t.co/bJCqbw6H61

[Yohan Lee @yohan0405]

The challenge asks what gets better when people and agents use a web app together. This is my answer.

Official: https://t.co/MvIBYZKcu2
Code (MIT): https://t.co/S8C0yDGz4E

What family-travel constraint would you test first?
#WebMCP

[Yohan Lee @yohan0405]

Built Model Relay for Codex: smaller models for bounded work, reusable task owners for revisions, and explicit verification before release. No new API keys or background services. v0.1 preview; live cost/quality benchmarks are still pending.
https://t.co/2U9idGW3qB

[Dilum Sanjaya @DilumSanjaya]

Asked GPT-6 Astra to create a highly detailed, interactive visualization of a V8 engine

The result is pretty impressive

[Gemini Notebook @Gemini_Notebook]

Yes, the rumors are true: The more powerful Notebook experience is now 100% rolled out to ALL Pro users.

You might be asking... How exactly do I use it? Here are a few queries to try (ofc dependent on your sources).

Hot Tip: bookmark this post! Free users too 😉

For Learning:

"Create 5 quizzes of ascending difficulty focusing on a different topic from my sources"

"Turn my raw lecture notes into a formatted PDF study guide complete with a 5-question practice quiz."

"Add all of the URLs from this class syllabus as sources"

For Business:

"Calculate revenue growth across each product line and plot the trends"

"Read our marketing brief, search the web for competitors, and recommend how we should tune our messaging"

For Personal:

"I uploaded a bunch of receipts from our kitchen renovation, make a spreadsheet to track all the work and costs"

"Help me analyze this property purchase. Look up recent market performance in the neighborhood, predict trends, and make a spreadsheet to model all the data you find"

[Thariq @trq212]

a skill people at Anthropic have been using a lot recently: ELI5

/eli5 <what you want explained>

"explain like I'm someone who knows nothing about this topic, using a HTML artifact with big pictures and few words"

[Alex Prompter @alex_prompter]

Y Combinator CEO, Garry Tan, took the stage for 42 minutes at Startup School 2026 and explained how to build your own personal AGI better than any paid AI course.

This is what he told the room:

1. The leverage is in your context, not the model.
Tan watches hundreds of founders use identical models every batch.
"There are 2x people and there are 100x people who are using the same Claude. Same weights, same context window size, same API. But the leverage is not in the weights."
The gap between users is now bigger than the gap between models.

2. One person's output went up 400x.
In 2013 Tan shipped maybe 14 useful lines of code a day as a YC partner, dead on the median for programmer productivity.
"I did the math on my output, and I'm at about 400x what I did in 2013."

3. Agents run on a different working memory.
Humans hold 7 things in their head at once. Every org chart and checklist ever built is a patch for that limit.
"An AI agent holds a million tokens. That's about a thousand pages. Three Harry Potter books sitting open on its head all at once."
You're still running your week on tools built for the 7-digit brain.

4. Markdown is code now.
Tan's stack is mostly skill files: pages of plain English an agent can execute.
"If you can write clear instructions in English, you're a programmer. The compiler is a language model."
At YC, finance and events staff who never opened a terminal are building automations.

5. Your history is your moat.
Tan's agent runs on a personal wiki: about 220,000 markdown pages covering 25 years of email, meetings, notes and decisions.
"When my agent does anything, it does knowing everything I know. And that's the difference between an assistant and a colleague."
No frontier model has your context. That's the one asset nobody can replicate.

6. Never do one-off work.
Most people run a task with an agent, close the window and throw the learning away. Tan ends every task by having the agent turn what it did into a reusable skill file.
"If you have to ask for something twice, you failed."
Captured skills compound daily. Amnesia resets you to zero every morning.

7. Own your skill files before your employer does.
A skill file is your judgment, extracted and executable. The only question is who controls it.
"Own your skills because if you don't, your job becomes a skill file."
Files in your repo compound your career. Files in the company's repo run your judgment without you.

Watch it, then read the step-by-step guide on becoming an AI engineer.

[Yohan Lee @yohan0405]

I just reserved my Cloudflare Wallet tag: https://t.co/cP4ep6SQxS.

Reserve yours now at https://t.co/jf4yjsGp6b

[Yohan Lee @yohan0405]

Voice coding works. Not because the agent understood one sentence — because I stayed in it for 52 utterances, said no more than yes, and checked the result.

The work moved from typing to deciding.

Full writeup: https://t.co/8QxR6KBARK

[Yohan Lee @yohan0405]

I built an app by talking to a coding agent.
85 voice turns. I kept the raw transcript.

The interesting part wasn't the app. It was reading back
how badly I communicate without a backspace key.

[Yohan Lee @yohan0405]

And the honest ending: my very last message in the log is me failing to install my own app on my own phone.

Nothing about voice or agents removes the last dumb step.