Me: Is this correct?
AI: Yes, absolutely.
Me: Are you sure?
AI: Apologies, you’re right. It’s incorrect.
Me: So which is it?
AI: Yes, absolutely.
We built a whole project around this conversation.
Imagine you grew up in the UK. You tell an AI you’re gutted about the match, and it explains that a fish has been cleaned for cooking.
You’d fix that instantly. The AI never will, unless someone tells it.
Now picture everyone who knows something a model gets wrong: a dialect, a trade, a local fact, a line of code.
That’s the vision behind Skillgraft. AI that gets better because people taught it, not just because someone bought more GPUs. The Floor is where it starts, and accepted work is rewarded in $SGRAFT.
Hot take: the next leap in AI won’t come only from more compute.
It will come from people who know things. The local who knows the slang. The developer who knows that function was never real.
That know how lives in people. We’re building a place to put it to work.
Think about the last time you corrected a chatbot.
You spotted the mistake. You explained it. Maybe you explained it twice because it argued back. Then you got your answer and moved on.
That was skilled work. And it vanished into a chat window where nobody will ever see it again.
Skillgraft flips that. On The Floor, your correction is the point. Find the slip, explain why, clear the 70% bar, and the accepted task is rewarded in $SGRAFT.
Same skill. Different ending.
Confession after building in Web3 since 2021: an AI once handed me a function that does not exist. It looked right. I trusted it way too long.
That’s a big part of why Skillgraft exists. AI doesn’t need more confidence. It needs people who say “No, and here’s why.”
Have you ever had a chat with any AI, asking for advice, only to realize what you were told was wrong?
How is that even possible, right? AI shouldn't make mistakes. That's why you trust it.
Well, not yet.
The agent economy still needs human judgement.
We're building that layer.
What would you teach it?
Powered by $SGRAFT, rewarding early adopters who are helping to facilitate that change.
ATMkvkBWRYozKPwSYj15HDu6ErRP9PjEPjkoNCRXpump
Thank you for the comment.
So far we have locked 1% of the supply for 12 months and used all creator fees to buy back more.
That is evidence of our commitment to building something worth discovering, not paying to look established.
Building a platform where people can actually contribute, complete assignments, and create value takes considerably longer than a quick pump.
DexScreener for $SGRAFT could come at any time, even though we feel that a working product matters more.
Buybacks will continue at randomized intervals.
Instead of frequent small buys, we will switch to larger purchases with the accumulated creator fees.
To clarify:
3% of $SGRAFT was bought at launch.
1% of that was immediately locked. 12 months lock, reserved for the team.
The remaining 2% goes to rewarding Skillgraft users.
Another 1% has been bought back, so the total held for users currently amounts to 3%.
We're open to receiving feedback about using the dev fees more efficiently.
More buybacks? Burn some?
Tell us what you'd like to be the future of Skillgraft.
1% supply successfully bought back.
All $SGRAFT held is reserved for rewarding early adopters (3% total so far).
The team's locked 1% will remain secure on Streamflow for the next 12 months.
The AI agent economy needs humans who check it.
skillgraft is building toward that layer.
$SGRAFT is being designed for member tools and program access.
Discover more:
https://t.co/lzuaaqhE2S
Paying Dex is not a matter of having the means to do it or not.
We chose to use all creator fees to buy back $SGRAFT as a way to signal our long term commitment (along with the 1% lock of the team allocation).
Many teams pay dex on the first day, but then disappear on the second day.
Skillgraft has been in development for a long time. It's not a quick cashgrab. That should make you feel confident in the team's long term commitment.
Coming soon: "Teach the AI Slang"
The machine presents a phrase such as "bro is cooked" and asks you to explain its actual meaning in context.
The new assignment includes internet slang, sarcasm, local expressions, cultural references, and regional language differences.
AI can give you the right answer and still sound completely wrong.
Robotic or confusing.
What if you could teach it how to sound more human and get rewarded for it?
A new type of assignment is coming to Skillgraft.
Your words, your judgment.
Your $SGRAFT.