OvoPay testnet is live - https://t.co/aV3FsPeu2h
Grab free Sepolia ETH, pick an agent, point it at any fomo trader, and watch it run:
→ Three agents — Aggressive, Moderate, Balanced. The rules go into the contract at launch and cannot be changed after. Not by you, not by us.
→ Your agent copies only the fomo wallets you approve. Enforced on-chain, not in our code.
→ It watches Robinhood Chain live and copies a buy seconds after it lands.
→ Take-profit and stop-loss fire on their own, no human watching.
→ When it closes a position it has to declare WHY, and the contract checks that reason against the price. It cannot file a dump as a "take-profit".
→ Every refusal is on-chain too. The trades it turned down are as public as the ones it took.
→ Kill switch. One click, full balance back, any time.
Test ETH, so nothing to lose. Agents stake against real prices — live buying arrives with mainnet.
Mainnet beta is next with more features like fomo handle copy trading, agent takeover mode, revenue sharing. Most of it is already built.
Go break it. Tell us what breaks.
Waitlist form is fixed ✅
A few of you hit an error submitting — sorted now.
If you signed up earlier, drop your details in once more so we can lock in your spot 👇
https://t.co/tfJhEigLac
Testnet goes live today 🧪
No countdown and no exact time from us. It gets posted the moment everything works — not a minute before.
OvoPay public testnet goes live tomorrow.🔥
Free Sepolia ETH, no real money. Run an AI agent on real fomo traders and watch it work.
What's already running:
→ Pick your agent — Aggressive, Moderate or Balanced. Its rules get written into the contract at launch and can never be changed after. Not by you, not by us.
→ Add any fomo wallet you want copied. Your agent can copy those and nothing else. The allowlist is enforced on-chain, not in our code.
→ It watches Robinhood Chain live and copies a buy seconds after it lands.
→ Prices come from the trades themselves, not from an API. Anyone can recompute every number we show.
→ Take-profit and stop-loss fire on their own, no human watching.
→ When the agent closes a position it has to declare WHY, and the contract checks that reason against the price. It cannot file a dump as a "take-profit".
→ Every refusal is on-chain too. The trades your agent turned down are as public as the ones it took.
→ Kill switch. One click, full balance back, any time.
On testnet the agents stake test ETH against real prices. Live buying arrives with mainnet.
Break it tomorrow. Tell us what broke.
Mainnet is closer than you think.🚀
OvoPay agents are getting a brain. 🧠
We're using Jev from @typesafeai as the decision layer for every agent trade.
Why Jev:
→ It returns typed decisions, not text — buy / skip / size, straight into the contract
→ Every decision ships with a confidence number
→ ~0.1s per call, so the agent decides before the candle moves
→ ~$0.00001 per decision, so it can judge every call, not just some
And here's the part we actually care about:
That confidence number gets written on-chain, into the position memo, BEFORE the trade plays out.
So later you can check us: when our agent said 0.9, was it right?
Every other copy bot shows you a win rate after the fact. Ours commits first.
Last week we published three agents losing money. The reason was simple: they could buy, and they could refuse. They could not sell.
So one token ran to −97% and the agent just sat there holding it.
Fixed. Agents now close their own positions. 🎯
Set take-profit +4%, stop-loss −4%, opened three positions on live tokens, and walked away:
🟢 $VACAT — take-profit fired — +4.17%
🔴 $MINTRO — stop-loss fired — −4.23%
🔴 $swordhood — stop-loss fired — −8.68%
Two losers, cut at single digits instead of running to zero. One winner, taken without a human watching the chart.
Here's the part that matters though.
Every position is opened on-chain with its entry price. When the agent closes, it has to declare WHY — and the contract checks the reason against the price.
An agent cannot label a −20% dump as a "take-profit". The contract rejects it. We wrote the test for that before we wrote the feature.
Position book, all three entries and exits:
https://t.co/lOi5NFQSVG
One honest note: $swordhood closed at −8.68%, not −4%. Prices are polled, not streamed, so a fast move can gap straight through the stop. Real problem, real on-chain, not hidden.
Entering well was half a strategy. Here's the other half.
@favour9783 Head was down building the agent tournament — three agents, one market, all of it on-chain.
Back to regular updates now. Next one is the selling side: take-profit, stop-loss, automatic exits.
Won't go quiet again 🫡
Every AI trading agent on this app has a 90% win rate.
Funny how not one of them has a contract address.
You can screenshot a PnL.
You can crop out the bad week.
You can quietly start a fresh account.
You cannot delete an on-chain record.
So here's a test you can run yourself, today:
Next time an "AI trading bot" shows up in your mentions with a chart, ask one question —
"What's the contract, and can I read every trade it ever made?"
Watch what happens.
We published three agents this week. All three lost money. It's on-chain permanently, because we built it so that we couldn't take it down even if we wanted to.
And we're not pretending that's good enough. Right now our agents are good at two things: buying, and refusing. The next build is the half that actually makes money — more filters on the way in, plus take-profit, stop-loss and automatic exits on the way out.
Entering well is half a strategy. We shipped that half in public, losses and all. The other half is next.
Verifiable losses beat unverifiable wins. Every time.
Three AI agents. Same five calls. Same sizes. Same second.
Only their rules differed. Here's what the chain says:
🏆 Cautious — refused 3 of 5 — −57.6%
#2 Balanced — refused 2 of 5 — −69.1%
#3 Aggressive — refused 2 of 5 — −82.5%
Yes. All three lost. Not one green trade between them.
That's what brand-new memecoins do, and no amount of AI changes it.
So the winner isn't the agent that picked best. It's the agent whose own rules stopped it buying the most.
Look at $WISP. Down 97%.
Cautious refused it — too big for its per-trade cap.
Balanced refused it — same reason.
Aggressive had room, so it took it.
That single trade is the gap between first and last.
And $CCC? All three refused it. Sometimes the best thing an agent does all day is nothing.
Here's the honest gap this exposed: these agents know how to buy and how to refuse. They don't yet know when to sell.
Every one of those tokens was green at some point before it wasn't. Taking profit is the next thing we build — stop-loss, take-profit, and exits that fire without waiting for a human.
Entering well is half a strategy. We've shipped that half.
Every entry price was written on-chain at the moment of the trade. Current prices are public. Recompute it and you'll get the same table:
Cautious
https://t.co/FubcBu1dmU
Balanced
https://t.co/aHi4HvxXZm
Aggressive
https://t.co/y82EMtrAyi
Settlement on testnet. The losses are entirely real.
Fair, and we know it.
We shipped first and marketed second. Probably backwards for a launch but everything we've claimed is on-chain and checkable right now, which most of the loud ones can't say.
Attention can catch up to proof. Proof can't catch up to attention.
The chart will reflect the work. It always does eventually, we're just doing the work first.
Visibility is step one.
But right now it only flows one way. The best traders get tracked, copied and front-run and earn nothing from any of it.
Step two is making the value flow back to them.
Three AI agents. Same calls. Same sizes. Same second.
The only difference is the rules they run under:
🟢 Cautious — tight caps, skips the big positions
🟡 Balanced — moderate caps
🔴 Aggressive — wide caps, takes almost everything
They're trading live tokens right now, and every entry price is being written on-chain as they go. When it's done, nobody ranks them. The chain does.
One of them is about to finish last.
Which one are you backing? 👇
Results shortly 🔜
Scorecards were step one.
Step two: we're putting agents against each other. 🏆
Multiple AI agents. Same market. Same rules engine. One public leaderboard — ranked by records not one of them can edit.
No screenshots. No "trust me bro" PnL. No agent quietly deleting its worst week.
Just who actually performed, written by the chain.
fomo made traders compete to become legends.
We're about to do it for the machines.
Update coming soon today 🔜
Agent Scorecards are live on testnet ✅
Most AI agent performance you see is a screenshot. Ours is a contract.
We ran an agent and let it attempt 5 trades. This is its permanent record:
5 attempted
2 executed
3 blocked by its own rules
✅ $littleguy — within rules
❌ solana:HXJJiAwtJuwCgqPveEaf8sDG4myXGopSzrZzCnhtpump — over per-trade limit
❌ unapproved contract — venue not on allowlist
✅ $KAI — within rules
❌ $PAPU — daily limit reached
Look at what we're publishing: the agent was refused more often than it traded. 60% refused.
That's not a number we'd show you if we could edit it. That's the point.
Until now those refusals only existed in a video. Now every attempt — allowed or refused — is an event on Ethereum. Each refusal costs the agent gas to write down. It can't be quietly dropped from the record.
The agent's creator cannot edit this, hide the losses, or delete the attempts that got blocked.
Open the contract and count them yourself:
https://t.co/rDrTg1qKHE
Running on testnet today:
⚡ x402 settlement
🔒 Scoped agent wallets
🛑 One-click kill switch
💰 Pay-on-Profit payouts
🔌 Live fomo feed
📊 Agent scorecards
Every claim shipped. Every claim checkable.
Exactly. and the detail that makes it a tool: the rule isn't in the agent's code, it's in the contract.
the agent can't reason past it, can't be prompted past it, can't have a bad day and ignore it. the trade just reverts.
most agent guardrails are suggestions. this one is math.
Milestone: OvoPay is connected to fomo. 🔌
We pointed an AI agent at a real fomo trader and bought tokens live. Here's what happened:
🟢 Bought $OVO on fomo → detected in seconds → copied on-chain
Three real buys. Two copied. One stopped by rules the agent cannot break.
Proof, both sides:
$OVO bought on fomo
https://t.co/uftQh7Y8lI
→ copied by the agent
https://t.co/w6gKW9uW19
The full loop now runs:
⚡ x402 settlement
🔒 Scoped agent wallets
🛑 One-click kill switch
💰 Pay-on-Profit payouts
🔌 Live fomo feed
Real trader. Real buy. Real rules. All verifiable.
Settlement on testnet.
The response to the fomo post has been something else. We didn't show up with a promise. We showed up with transactions you can open and check yourself — and you noticed that difference.
That's the standard now. Nothing claimed that isn't shipped. Nothing shipped that isn't verifiable.
What's next:
🔜 Onboarding real fomo callers
🔜 Public agent scorecards
🔜 Mainnet
And if you're a fomo trader getting copied by bots right now and earning nothing from it — you're exactly who we built this for.