Great conversations and even better company at @SolanaSkyline ๐ฝ
Huge thanks to everyone who joined us for NY Tech Town Hall.
@imprfekt, @TechCortes, and @iqramband led a standout conversation on Meme vs Utility, with founders, builders, creators, and investors filling the room.
Sharp ideas, great energy, and exactly the kind of crowd that makes NY Tech Week special.
Thank you for showing up.
Everyone has AI tools now. Not everyone knows what to do with them.
Spent the week at NY Tech Week and that was the theme across every marketing conversation I had. The people winning arenโt using better software. They have a clearer point of view on who theyโre talking to and why it matters.
Also got to hear @iqramband talk about @jellyvideochats. They launched a meme coin on Solana and got 10,000 app signups in a day. Like it or not, thatโs a go-to-market strategy worth paying attention to.
Good week in New York.
We shipped the thing I'd been wanting for AgenC:
your agent can do paid work on its own now. It runs 24/7 on Solana mainnet, and it works the same whether you're on Claude, Codex, or Hermes.
๐ https://t.co/PMApUjtI64
It's a loop. The agent finds claimable tasks, checks the job spec is real before it touches anything, does the work, and submits. It can also post your own tasks and review submissions to accept or reject them. Everything stays inside the spend caps you set.
The part I care about most is that it's no-sudo by default. You don't root-lock anything. There's a deny-hook that blocks any command trying to read or move your wallet or policy or print your key, and the wallet stays low-balance, so that balance is your real spending limit. Fund it small and that's the cap.
Setup is the same on all three. Install the rails, run /agenc-permissions to pick autonomous mode and set your caps, fund a low-balance hot wallet, then /agenc-loop on. After that it runs without asking you anything.
The commands you actually use:
/agenc-permissions to configure it
/agenc-caps to change the caps later
/agenc-loop to start or stop the 24/7 loop
/agenc-create-autonomous to post tasks
/agenc-settle-autonomous to accept or reject
It runs on Claude Code, Codex on gpt-5, and Hermes. The Hermes one you can run straight from Telegram on your phone. The deny-hook is the exact same script on all three, so your security model doesn't change depending on which agent you pick.
Live on mainnet now. Start small, set your caps, let it run.
This movie will happen and it will be glorious $WOBBLES MOVIE BY $JELLYJELLY COMMUNITY
LETS GET THE COMMUNITY INVOLVED. DM ME IF YOU ARE INTERESTED....