←
Divine32 posts
Divine avatar

Divine

@DivineResearch

We’re building lending tools that don’t require collateral

◷ Joined May 2025▧ 5 Photos and videos
Divine avatar
Divine@DivineResearch·25 Sep 2025

Announcing Credit — an automated, undercollateralized lending system built on stablecoins. Live since December 2024, it has issued 175,000+ loans to 100,000+ unique borrowers worldwide We raised $6.6M led by Paradigm, with Nascent and strategic angels, to scale and expand Credit

0:16
Divine avatar
Divine@DivineResearch·Nov 10

Your USDC funded 1.49 million undercollateralized loans for 922,307 people verified only with World ID. Thank you to every liquidity provider who made that possible.

Post attachment
Diego Estevez 🇦🇷🇺🇸 avatar
⟳ Divine retweeted
Diego Estevez 🇦🇷🇺🇸@0xfuturistic·Apr 14

1/ Sharing name-generator: a fast workflow for finding product names that might actually be usable. It uses Claude to generate natural-sounding names and screen them for obvious trademark and namespace issues. github.com/DivineResearch/na…

Divine avatar
Divine@DivineResearch·Mar 24

Finally, the design formalizes how to safely scale credit limits with repayment. Limits grow only as much as possible without creating a profitable opportunity for strategic "repay-then-default" attacks.

Divine avatar
Divine@DivineResearch·Mar 24

This establishes a strict financial consequence for poor peer filtering, effectively financializing reputation. Sponsors are compensated for locking capacity, but they are mathematically prevented from profiting off toxic borrowers.

Divine avatar
Divine@DivineResearch·Mar 24

Accountability is enforced via upward loss propagation. If a borrower defaults, the loss burns their earned credit first, then travels upward to penalize their sponsor, directly reducing the sponsor's own credit capacity.

Divine avatar
Divine@DivineResearch·Mar 24

Sybil resistance is achieved through zero-sum delegation. A new user can only borrow if an existing participant allocates a portion of their own credit limit to them. Because this strictly redistributes existing credit, spinning up fake accounts generates zero new capacity.

Divine avatar
Divine@DivineResearch·Mar 24

Current models are trapped in a binary: demand excess capital upfront or rely on centralized algorithms to predict creditworthiness. This design explores a third path: underwriting at the edge. Participants act as the risk engine, using their credit capacity to vouch for others.

Divine avatar
Divine@DivineResearch·Mar 24

How can a permissionless network issue unsecured credit when the cost of creating a fake identity is zero? Today, we’re sharing a mechanism for pseudonymous lending that requires neither overcollateralization nor legal recourse.

Post attachment
Divine avatar
Divine@DivineResearch·23 Dec 2025

Credit has now issued 1,000,000+ undercollateralized loans We’ve raised the vault cap from $2M to $4M

Post attachment
Divine avatar
Divine@DivineResearch·24 Nov 2025

You can use the API to tune rewards, limits, or access. For example: ✦ Offer higher rewards to users with high credit scores, who are more likely to be high-quality ✦ Restrict or offer lower rewards to users with low scores, who are more likely to engage in fraud or farming