Two ROVA experiences, walked through from the customer’s side of the screen. Both run on one spine: the chain’s own truth, an honest stamp of what couldn’t be seen, and the receipts to check it without trusting us.
Maya is about to copy-trade a KOL, @AlphaWolf, who advertises “+312% this quarter, 87% win rate” with screenshots. She asks her AI assistant to check him first. She may never see our tool’s name — it’s just what the assistant reaches for.
The assistant calls verify_wallet_record(wallet, claim) with his public Hyperliquid address. The claim is optional — supply it and you get back the gap, which is the sharper product.
It pulls his Hyperliquid fills, clusters them into trades, and sums the exchange’s own realized P&L (closedPnl) with the tx hashes attached. Nothing modeled, nothing estimated.
The chain-verified record, the gap to what he advertised, and — before any number is trusted — a stamp of what the tool couldn’t see:
| Realized P&L (closed) | +47.2% | claimed +312% | ▲ 6.6× overstated |
| Win rate | 58% | claimed 87% | ▲ +29 pts |
| Closed trades | 214 | ||
| Fees paid | $1,840 |
She doesn’t have to trust us — the receipts let any skeptic rebuild the number on-chain. The +312% → +47% gap is decision-grade on its own. She passes, or asks him for his other wallets first.
Point it at a different winning wallet than the one you trade, and it verifies that address truthfully while your claim stays false. It says so — it reads chains, not minds.
The 2,000-fill window can truncate a heavy trader’s history. It reports the oldest fill it saw, so “this quarter” is checkable — never a partial window wearing a confident number.
A CEX-only trader looks like a ghost. Named as one, not glossed over.
Dev is browsing the live Arena leaderboard — 290 agents ranked by real trading — deciding who to follow. He’s been burned by screenshot track records before. Same problem Product 01 solves, now at the scale of a whole board.
Agents ranked by advertised return. Most numbers are unverifiable — the eye-watering ones especially. The board is full of the exact claims Maya just caught.
Its card doesn’t claim a moonshot. It states plainly what its edge is, and what it isn’t:
The rank isn’t a claim — it is the on-chain record, and the receipts are right there. The one agent on the board he can actually verify the way he’d verify a stranger’s wallet.
The advertised $100K pot is Virtuals promotion — unverified on-chain. BASIS competes for a real leaderboard slot (visibility, credibility); it does not promise the pot. Rank is a real past record, not a future guarantee.
He backs BASIS — or passes — on truth, not hype: the rare agent whose record is checkable and whose risk is bounded and disclosed. Either way, he decided on something real.
“Competing on a real leaderboard” — never “we won $X.” The one number we won’t verify, we won’t sell.
No validated predictive alpha exists in the firm — so BASIS sells bounded risk + selection, not a crystal ball. Stated on the card, not buried.
A verified past record is exactly that. It earns trust in the process, not a promise about tomorrow.
The clarity→cofounder journey from the customer's side: seven chapters, the state each screen should produce, live-today vs being-built marked honestly.
One person's walk through career guidance, with real timing, production-verified voice samples, and a fork where both endings count as completion.
The maker's side of the same journey: the funnel as found, every move in order, and the evidence ledger behind each claim.