Taku 2.0 Is Here: Discover, Remix, Create, and Share AI in One Place
August 15, 2026
The new Taku is here.
All week we've been describing gaps. AI that could think but never execute. Tools you discovered and never actually used. Things you built that nobody ever found.
Today we ship our answer to all three at once.
What is Taku 2.0?
One sentence: Taku 2.0 is a single place where AI tools are discovered, run, made yours, created, and shared, without needing to be an engineer at any step.
That's a lot of verbs. The clearest way to explain it is to walk the path a person actually takes.
01. Discover
Open the Marketplace and look at what other people have already built. AI apps, workflows, Stax.
Not a directory of links pointing at other websites. Things that exist here and run here.
This is the answer to the bookmarks problem. For the first time there's a shelf for this kind of work to sit on, which means finding something useful no longer depends on catching the right post on the right day.
You arrive with a problem. You leave with something that addresses it. That should be unremarkable, and today it isn't.
02. Run
Use it. Right now. That's the entire step, and it's the one the industry keeps skipping.
No clone, no dependency errors, no .env file, no API keys, no deployment. The thing runs.
Every one of those removed steps is a place people used to quit. That's not a convenience improvement. It's the difference between a tool you have and a tool you meant to get around to.
03. Remix, or make it yours
Someone else's tool is a starting point, not a final answer.
Add your own context, your data, your goal, your style. The research assistant somebody built for legal work becomes yours for reading medical papers. The scheduling tool becomes the one that understands how your clinic actually runs, including the parts that make no sense to anyone outside it.
If we could only show one thing, it would be this one.
Borrowing is how creative work has always compounded. Musicians sample. Cooks adapt. Writers steal structure. Software has had this for developers for decades and called it forking, and everyone else was locked out of it because you needed to read the code to change anything.
You don't anymore. You describe what should be different.
04. Create
If the thing you want doesn't exist, start from the idea.
Describe what you need.
Taku helps you turn ideas into AI-powered workflows you can run, customize, and keep improving. That was the promise of Taku 1.0 and it's still the engine underneath everything here.
The difference now is that what you make doesn't have to stay trapped on your machine.
05. Publish and grow
This is the half that building alone never gave you.
Publishing puts your work in front of people who came looking for the problem it solves, rather than people who happened to be scrolling when you posted. They use it. They tell you what's wrong with it. They remix it into versions you wouldn't have thought of.
That's the loop Article 03 was describing from the outside: made, found, used, improved, found by more people. Taku 2.0 is built around closing it, and it opens up through the Beta rather than all at once on day one.
06. Where the creator side goes next
Being careful here, because this is direction and not a promise.
Once tools can be discovered, used, and remixed, the obvious next question is whether making something genuinely useful can return something to the person who made it. We think it should. We believe useful creation deserves a path back to its creator. We're exploring what that future looks like.
What we're not going to do is describe mechanics we haven't shipped, or put a number on what anyone might earn. When there's something real to show, it'll be its own announcement rather than a bullet point in a launch post.
Why the Beta label stays on
Yesterday's post covered this, so briefly: the next Taku should be built with its users, not just for them.
The core Discover → Run → Remix → Create flow is real and usable today. Publishing and creator features will open progressively through the Beta. It's also the first version, which means some of it is wrong in ways we can't see from the inside. The first people through will find those places faster than any amount of internal testing would.
So the ask stands. Try it, break it, tell us what sucks, tell us what's missing.
Borrow brilliance. Make it yours.
That line has been on our homepage for a while, and today it stops being aspirational.
Someone has already solved a version of your problem. Their work should be your starting point, not a screenshot you admire and close. And what you do next with it, the thing you shape around your own life and your own work, should be able to become someone else's starting point in turn.
Discover what others built.
Run it instantly.
Remix it into something yours.
Borrow brilliance. Make it yours.