ımyo
pre-launch · MMXXVI
№ 13 · Product

The credit terminal.

two buttons · one screen · zero theatre

Two buttons. One screen.
The AI handles the rest.

ımyo is what institutional DeFi lending looks like when you remove everything that doesn't need to be there. No governance theatre. No yield-token theatre. No fifty-asset menus. No five-tab dashboards. A treasurer logs in, sees one balance and one risk gauge, presses one of two buttons, and the protocol does the rest.

The first generation of DeFi gave operators every knob, every market, every primitive — and asked them to assemble the venue themselves. That worked for crypto-native users who wanted that surface. It does not work for the people writing the next $16 trillion of allocation memos. The institutional question is not "how many features does it have", it is "how few decisions does it ask me to make in order to be safe". The answer, on imyo, is three: what to lend, what to borrow, and whether to keep the position open.

Aave is the Bank of DeFi. ımyo is the credit terminal. positioning · 異名
№ 13-A

Six lines, in lieu of a product tour.

the bet, in plain text
surfaceTwo buttons.
layoutOne screen.
settingsZero knobs.
governanceZero theatre.
cross-chainZero bridges.
tokenomicsZero token.

Everything below is the long version of that.

№ 13-B

What we refuse to ship.

the cut list

Most of the surface area of a legacy lending protocol is not the product. It is the residue of five years of governance decisions, token incentives, and feature pile-ups that no one had a mandate to clean up. ımyo ships without them — not as a roadmap milestone, but as a founding constraint.

Carried by Aave / Compound / Morpho
Why imyo does not
Safety module · token-staking insurance
Dilutive, opaque, and dependent on token price to be solvent. Replaced by AI-attested live risk plus a real-revenue reserve.
40–60 listed assets · e-Mode · isolation mode
Curated to a small Tier-1 / Tier-2 / Tier-3 set. Quality of collateral, not quantity. The AI onboards new assets in days, not by community vote.
Stable and variable borrow rates
One rate per asset. Priced by the AI against macro signals. A treasurer should not need to choose a curve.
Permissionless market creation
Open creation is a security surface, not a feature, for institutional capital. AI curation absorbs the upside; the user never sees the firehose.
Vault curators · vault managers · vault layers
Lego-towers of intermediaries that obscure what the user is actually exposed to. One protocol surface, one accountable party.
Protocol-native stablecoin
Distraction. Use real stablecoins — USDC, PYUSD, USDY, BUIDL. Don't compete with your own collateral set.
DAO forums · Snapshot · voting UI
Policy lives with the 7-of-11 Guardian under AI attestation. Users never see governance. The product is the product.
Token incentives · points programmes · airdrop farming
Mercenary deposits are not deposits. Yield is real or it isn't. No bribe layer.
Cross-chain "portal" bridges
The single largest failure surface in DeFi history. Architecturally banned. Sui and Ethereum stay siloed. See architecture.
Multiple protocol versions (V2 / V3 / V4 churn)
One protocol. Upgrades happen in-place under continuous AI audit. There is never a second deployment to migrate to.
№ 13-C

Three jobs.

the entire product

The product does exactly three things. Anything that is not one of these three things is not in the product.

01 · Lend

Deposit. Earn.

One input — amount, asset. One button. The AI routes capital to the right internal pool, prices the rate against macro signals, and rebalances inside the silo as conditions change. The treasurer sees a balance and an APY. That is the entire experience.

02 · Borrow

Pledge. Draw.

One input — amount, collateral. One forward-looking risk reading before the user signs. The AI sets the LTV per asset, per duration, in real time. Predictive liquidation buffer rebalances 15–30 minutes ahead of a hard event. The user is told what is happening; the user is not asked to manage it.

03 · Manage

Watch. Close.

One position screen. Live LTV, live AI signal, live rate. One button to close. One button to rebalance. One download for the audit-ready statement that goes straight into SAP or Oracle. No drawer of secondary actions. No "advanced" tab.

№ 13-D

Four screens.

the entire surface

That is also the count of the application. Four screens, end to end. If a feature does not fit on one of these four, it is the wrong feature. The mockups below are deliberately low-fidelity: the institutional reading of a product is not the screenshot, it is the count of the things the user has to know.

/app · portfolio
balance                $ 1,247,300.12
risk gauge             ● ○ ○ ○ ○
AI signal              stable

open positions                      3
last statement             apr 2026 ↓

       [  LEND  ]      [ BORROW ]
One number, one gauge, one signal, two buttons. The dashboard exists to confirm the position is fine — and to take the user to one of the two flows.
/app/lend
asset                  USDC ▾
amount                 $ ________

current rate           4.81 % apy
AI signal              widening
silo                   Ethereum · Tier-1

           [  CONFIRM DEPOSIT  ]
Two inputs, one explanation, one button. The rate is what it is — no kink-model, no slope, no choice of curve. The AI prices it.
/app/borrow
collateral             BUIDL ▾
draw                   $ ________

LTV (AI-set)           88 %  →  82 % (30 min fwd)
forward risk           contained
predictive buffer      armed · 22 min

           [  CONFIRM BORROW  ]
Two inputs, one forward reading, one button. The user is shown what the AI sees in 30 minutes — and then it just does it.
/app/positions/p-0xa3
collateral    1,250,000 BUIDL    $ 1,250,000.00
debt                              $   980,000.00
live LTV                                 78.4 %
AI rating                          contained ●
next AI check                          14:32 UTC

   [ CLOSE ]    [ REBALANCE ]    [ STATEMENT ↓ ]
One screen per position. Three buttons. Statement downloads as a T+0 institutional report — SAP, Oracle, TaxBit-compatible.

The reporting page (/app/reports) is the fifth surface but is intentionally off the main flow — a treasurer goes there once a month, not every visit.

№ 13-E

Five engineering commitments.

enforced, not aspired

Each of the five is a constraint applied to every pull request. Anything that violates one of them does not ship, regardless of how attractive the feature.

  1. No screen has more than three primary actions.

    If a fourth is required, the screen splits. Three is what a human reading the page in three seconds can hold.

  2. No setting is exposed that the AI does not already solve.

    If we would ask the user to choose, that is a bug. The AI chooses. The user reviews and signs.

  3. No flow takes more than two clicks from connect-wallet to executed transaction.

    One click to open the flow, one to sign. Anything in between is friction the AI should have absorbed.

  4. Median time to first quote: under 200 ms.

    Quotes are simulated at the Cloudflare edge before the transaction reaches a chain. There is no spinner; there is the number.

  5. The CSS file is shorter than the marketing copy.

    Visual restraint as a hard, measured constraint. Refinement comes from typography and spacing, not from animation libraries.

№ 13-F

In one line.

the whole pitch

The credit terminal.

Aave is a Bloomberg terminal that grew barnacles · imyo is what is left when you saw the barnacles off
№ 13-G

Read next.

deeper
11 · vs Aave

The neutral comparison.

Where Aave is the correct answer, and where imyo is. No hype.

read 02 · Thesis

The institutional thesis.

Who is on the other side of these two buttons.

read 04 · Silo

Siloed liquidity.

Why "zero bridges" is a hard constraint, not a roadmap line.

read 05 · AI

The AI doing the work.

How the auditor and sentinel swarms keep the surface this small without losing the safety.

read