Let's see if GPT-4 can build ChatGPT in 10 prompts or less:
Note: I'm going to do this live, I have no idea if this is going to work. Wish me luck 🫡
PROMPT 1:
We're asking it to build a React component for a chatbot:
Here's the response, it looks pretty good!
PROMPT 2:
Now we'll ask it to write the HTML file that includes the React component, and style it. I asked it for dark mode–let's see how it does 😅
Here's the response:
As you can see, it's asking me to include an index.js file...but it hasn't yet told me what that is. Let's use a prompt on that!
Here's its response:
It's recommending I use the create-react-app to start the project, which makes me a bit nervous but we'll give it a shot!
I'm creating the React app now:
AND WE HAVE A CHATBOT!
It's client-side only for now, but you can send messages and everything :)
PROMPT 3:
Okay now we're cooking! We need a Flask server to handle the chat requests from the client. When it gets a chat request it should send it to GPT-4 for a completion. I'm literally going to paste in the GPT-4 API docs for reference.
Let's see what it can do!
Here's the response. Looks good!
I'm going to implement this (with a few minor modifications) and see what happens...
Our server is up and running!!!
Okay, we ran into a minor problem. Our server and client are hosted at different addresses—so when I make POST requests I get a Cross Origin error:
PROMPT 4:
Let's ask GPT-4 to fix the CORS error!
Here's the response. Seems reasonable! I'll report back...
Okay I went back and forth a bit on the CORS issue until it finally found the right modification for my app.py file. Not counting that as a prompt...it's sort of a side quest conversation. But...it fixed it!
Now, when I type a message it hits my chat server!
But there's a problem...when I send the request to GPT-4 it returns an error.
It looks like for some reason we're not actually sending the chat transcript to GPT-4. Not sure why...I'm going to step through the code
Okay so it looks like for some reason the client side isn't actually sending the chat transcript to the server.
I wonder why that is...
OMFG IT WORKS
For some reason it works on the second message, but not the first.
Going to have to ask GPT-4 why that is, and see if it can fix it...
PROMPT 5:
I asked it why the first message doesn't arrive and it figured it out!
It worked! It worked! It worked!
This was shockingly, crazily, incredibly easy.
I actually also learned something...I never would've used the create-react stuff on my own. I would've rolled it myself.
But it's quite good!
Now...let's see if I can get my ChatGPT clone to create another ChatGPT clone in 10 prompts or less... 👿
• • •
Missing some Tweet in this thread? You can try to
force a refresh
It’s a live collaborative document editor where humans and AI agents work together in the same doc. It's fast, free, and open source—available now at proofeditor.ai.
It’s built from the ground up for the kinds of documents agents are increasingly writing: bug reports, PRDs, implementation plans, research briefs, copy audits, strategy docs, memos, and proposals.
Why Proof?
When everyone on your team is working with agents, there's suddenly a ton of AI-generated text flying around—planning docs, strategy memos, session recaps. But the current process for collaborating and iterating on agent-generated writing is…weirdly primitive.
It mostly takes place in Markdown files on your laptop, which makes it reminiscent of document editing in 1999.
Proof lets you leave .md files behind.
What makes Proof different?
- Proof is agent-native: Anything you can do in Proof, your agent can do just as easily.
- Proof tracks provenance: A colored rail on the left side of every document tracks who wrote what. Green means human, Purple means AI.
- Proof is login-free and open source: This is because we want Proof to be your agent's favorite document editor.
Proof is the most collaborative product we’ve ever built @every. It has commits from 5 team members, both technical and non: @bran_don_gell, @kieranklaassen, @naveennaidu_m, @bigwilliestyle, @lucas__crespo.
It is such an immense pleasure to work with this team, and it makes me so excited to make this product open source.
Let’s work together to make Proof as good as it can be:
I just showed the team our 2026 roadmap. I got so psyched that I wanted to share it with you too.
Here's our full 2026 strategy.
Every is built around three things:
Ideas. Apps. Training.
Everything connects to that mission. And this year, everything we build will connect to each other.
We run on a loop:
- Live in the future together.
- Write what we see.
- Build what's missing.
- Teach what works.
Writing is the engine. Force yourself to articulate what you're seeing, and you make that future visible to everyone else. Once you've named it, it's obvious what to build next.
i've been playing with it for a week or so and it's already my go-to model. it's fast, agentic, extremely smart, and has great vibes.
some of my top use cases:
- it flagged every single time I sidestepped conflict in my meeting transcripts
- it spun up a bite‑size ML course that it pings me about every morning
- it found a stroller brand from one blurry photo
- it coded a new custom AI benchmark in record time
- X‑rayed an Annie Dillard classic and found writing tricks I’d never noticed before
- it even analyzed @every’s org chart to tell me what we’ll be good at shipping, and what our weaknesses are