Introducing CipherLend, Every lending app asks you to hand over your salary, income, debt and
savings before it'll tell you if you qualify.
I built CipherLend so it never has to ask.
A confidential loan-eligibility engine that computes entirely on encrypted
data, built on
@zama FHEVM.
#ZamaDeveloperProgram
The core problem: underwriting isn't a single fact you prove once.
It's live math across several private numbers at the same time debt vs. income, savings vs. a buffer, credit vs. a tier, recomputed per
application.
A ZK proof checks one fixed threshold. It can't do that.
Fully Homomorphic Encryption can.
CipherLend encrypts your financial profile in the browser, submits only
ciphertext on-chain, and runs the entire eligibility model, DTI checks,
risk tiering, loan sizing, interest tier as arithmetic directly on
encrypted values.
The lender gets exactly 4 outputs, and only if you authorize them:
✅ / ❌ Eligible
💰 Max loan amount
📊 Risk category
📈 Interest tier
They never get your salary. Never your debt. Never your savings. Never
your credit score. Not once, not ever.
Under the hood: FHEVM doesn't support ciphertext ÷ ciphertext division, so
DTI checks are rewritten as cross-multiplied comparisons:
Debt/income ≤ 0.40` becomes debt × 100 ≤ income × 40
Same decision, zero division, fully homomorphic.
Encrypted control flow can't branch normally, the EVM would need to know
the plaintext to pick an if / else path.
Every ciphertext has its own on-chain ACL. Raw profile data is permissioned
to the borrower and only the borrower forever. Decision outputs get
selectively shared through authorizeLender(), revocable at any time.
It's live on Sepolia real transactions, not a simulation:
Contract:
sepolia.etherscan.io/address…
Code:
github.com/Mani678/cipherlen…
Live app:
thecipherlend.vercel.app/
If FHE is the missing piece for confidential DeFi, lending is one
of the clearest places to prove it.
Prove you qualify. Reveal nothing else.
#ZamaDeveloperProgram