10 years building collaboration SaaS. Now trying to fix how humans work with AI.
I've been building collaboration tools for 10 years. Whiteboards. Canvases. Real-time spaces where teams get on the same page. Raised $6m from Zoom and other VCs, spent 5 years in Silicon Valley. Built a real product with real users.
One thing I learned: the best work happens when people align first. See the same picture. Agree on direction. Then execute.
Alignment before action. Every time.
Then AI happened.
I started using ChatGPT, Claude, all of them. And something felt off.
You type one sentence. AI builds something. Fast. Confident. Generic.
No alignment. No shared understanding. Just guessing.
I'd put in a good idea and get back average output. The better my idea, the worse it felt. AI kept flattening what made it mine.
I thought: we'd never run a meeting this way. We'd never start a project this way. Why do we work with AI this way?
What if we brought the whiteboard into AI?
That question became Rebrew.
How it works:
- You drop a raw idea. No prompt engineering.
- AI shows you 3 directions with real tradeoffs. You pick one.
- That becomes a canvas. Visual. Editable. You shape it with AI. Move things around. Ask AI to review. Challenge gaps. Like working with a collaborator, not a slot machine.
- When it's ready, one click turns it into a landing page, blog post, social content. All from the same source.
You align with AI before it builds. That's it.
Where we are:
Launched on Product Hunt today. Still early. Still an experiment.
https://www.producthunt.com/products/rebrew?utm_source=other&utm_medium=social
I don't know if this is the right interface. But I know the current one is broken.
Looking for honest feedback from people who build things.
What's missing? What feels off? What would you actually use this for?
Doing something with researchers and ai on epion. ai please check it out you will like it
Tried it and it looks nice. At least showing the canvas (no actions allowed) for free would be interesting. I cant really tell the advantage of asking chatgpt to give me 3 directions. Also a screenshot of a few canvases on the homepage? And to me it is not clear how these 3 directions represent alignment and understanding. Wouldn't it first need a bit of back and forth to get the idea across. The directions seemed interesting, but also too specific, guessed from too little context.
Thanks for trying it out.
You're right. The directions come too fast. Not enough context, feels like guessing.
We went this route to avoid the "20 questions" problem where AI keeps asking before doing anything. But the tradeoff is what you experienced. Directions that feel specific but disconnected from what you actually meant.
Adding files or more detail upfront helps. But that's not obvious in the current flow. We need to make that clearer. Or find a better balance between asking and assuming.
Showing example canvases on the homepage. Agreed. Would help people see what they're building toward.
Appreciate the detailed feedback. Genuinely helpful!