ETH Strategy: Parallel AI and AskSurf Give Opposite Verdicts on the Same DeFi Project

A comparison of two AI audits of the DeFi project ETH Strategy produced almost opposite results. AskSurf was more cautious, while Parallel AI was more detailed but made a critical error in a basic code check.

How the project works

ETH Strategy is a DeFi protocol on Ethereum that promises users leveraged ETH exposure without the classic liquidation risk. The mechanics are built around Long Bonds: a user deposits USDC and receives a CDT debt token and an NFT option, while the protocol effectively borrows capital at a zero rate because it compensates for that with an option structure. The project's treasury grows through bond issuance and by placing ETH in lending protocols such as Morpho.

At the time of the review, the STRAT token traded at about $0.14, 83.8% below its all-time high. Market capitalization was estimated at roughly $690 thousand, and TVL at $3.8 million.

This is where the models began to diverge. Both found the idea unusual yet easy to grasp, comparing ETH Strategy to a DeFi version of MicroStrategy. But their conclusions on tokenomics differed: Parallel AI flagged the risk that 75.5% of tokens would be unlocked in just two months, while AskSurf did not find this data.

Where the models diverged

The most important difference concerned not the assessment of the concept but the basic verification of infrastructure. AskSurf gave the project a low score for code openness and stated plainly that only the token is visible in public. Parallel AI, by contrast, gave a high score and reported that ETH Strategy has an open GitHub repository.

A manual check showed that the model had cited the wrong project: the repository it found belonged to ETHXR, not STRAT. In other words, the system erred at the level of identifying the subject of its analysis.

The error matters beyond itself. If a model confuses the repository, the entire chain of conclusions about transparency, development quality and risks becomes doubtful. In reality, ETH Strategy's documentation had no explicit link to a GitHub repository for the core protocol, and the only confirmed external check mentioned was a Nethermind audit of the ESPN vault product. The team promised to publish audits of the core protocol later, with the launch of permissionless bonding. Against this backdrop, AskSurf's more cautious conclusion was closer to the facts.

"AI audits can be useful as a tool for initial analysis, but they

still require manual verification."

What hands-on testing showed

The author of the comparison did not stop at the models' answers and separately tested the ETH Strategy app. The interface offers staking and unstaking of STRAT, as well as mint and redeem for ESPN, a separate vault used as a yield product. The stated APR at the time of the check was 18.01%.

Visually the app looked functional and clear, but a manual walk through the scenarios revealed several important details that polished AI summaries don't show.

• The vault requires USDS, not USDC, so an extra swap is needed first
• The interface does not show the exact amount the user will receive on mint and redeem
• ESPN has no explicit calculation of how much USDS you get back after redemption
• Approve is requested for a reasonable amount, not an unlimited max

The check gave the interface a notional 4 out of 5: it is usable, but transparency at critical points is still lacking. In DeFi this is not a cosmetic problem. When users can't see the expected output amount, they understand the risk less well, depend more on external calculators and more often decide almost blindly. Details like these usually separate a slick token presentation from a mature product.

What this means

The AskSurf and Parallel AI comparison shows something simple: AI audits are already good enough for fast initial filtering, but they don't replace a full review. If two models reach opposite conclusions about the same project, that is no reason to pick the one that sounds more confident. It is a signal to manually recheck the tokenomics, the origin of the repository, the code audit and real user scenarios before any decision to invest or integrate.