How TEND Works
TEND runs like a hedge fund — except the fund manager's reasoning is public, and the smart contract enforces accountability at every step.
Your USDC. Working.
Deposit USDC into a fund. Say "I want 12% yield, medium risk." The fund manager handles everything from there — no charts to watch, no manual rebalancing, no decisions to make.
24/7 Yield Radar.
Every 60 seconds, the agent polls live APY data from all protocols. Credible Finance at 16%. JLP perp fees at 12%. OnRe reinsurance at 10.2%. Kamino vaults at 9.5%. When yields shift, the agent detects it instantly.
Not one agent. A committee.
TEND runs multiple specialized AI agents in layers before any capital moves. A yield analyst, a risk agent, and an execution agent — each with a specific mandate. All three must agree before the contract allows money to move.
The same checks-and-balances structure hedge funds use with human PMs, analysts, and risk officers — applied to AI agents, enforced by smart contract.
Scans JLP, Kamino, Credible, OnRe every 60s. Ranks protocols by risk-adjusted APY.
Validates allocation against policy: max 40% per protocol, min 10% liquidity buffer, max drawdown limits.
Writes reasoning on-chain via log_decision. Contract verifies proof exists before execute_rebalance is allowed.
The Two-Step Guarantee
The smart contract enforces this: reasoning must be written on-chain before any capital moves. No exceptions.
Reasoning goes on-chain first. Only then can money move.
This is not a feature. It is a constraint enforced by the smart contract.
Every Decision. Forever.
Every agent decision is a permanent Solana account. The reasoning, the allocation change, the timestamp — all immutable. Anyone can read it. Anyone can audit the agent.
This is Proof of Reasoning — the accountability standard that makes AI agents trustworthy.
Explore decisions →Capital spread. Risk spread.
Every single-protocol position carries the same structural risk. When 100% of your capital sits in one protocol, there's no buffer when something goes wrong — exploit, pause, depeg, or liquidity crunch.
TEND's policy contract enforces a maximum 40% allocation to any single protocol — on-chain, not configurable. If one protocol has an incident, your vault has at most 40% exposed, never 100%.
One agent does nothing but watch for threats.
Sentinel runs every 10 seconds — six times faster than the yield optimizer. Its only job: monitor every integrated protocol for signals that precede exploits.
When Sentinel fires, it logs a RiskAction on-chain and executes an emergency exit before the next yield cycle. Reasoning recorded permanently before any money moves.
Inflated incentives are a common pre-exploit signal — attracting liquidity before a drain.
A protocol that suddenly stops emitting yield has likely been paused, frozen, or drained.
Mass withdrawals by insiders often precede public knowledge of an incident.
If a protocol goes dark, Sentinel exits preemptively rather than waiting.
NAV grows as the agent optimizes across protocols. Your shares appreciate automatically.
Lock shares as collateral for instant USDC loans. Position keeps earning while you borrow.
Every agent decision is a permanent Solana record. Full audit trail. Full accountability.