Modulus Labs

Modulus Labs was a company working at the intersection of artificial intelligence and blockchain, applying zero-knowledge cryptography — specifically zero-knowledge machine learning (zkML) — to make AI computation verifiable on-chain. Its stated aim was to let smart contract developers and on-chain services incorporate AI outputs without abandoning blockchain-level security guarantees: rather than trusting an opaque model running on a server, a counterparty could verify a cryptographic "proof of correctness" that a given model was executed as claimed, at lower cost than re-executing the computation. Company co-founder Daniel Shorr compared the resulting assurance to a verification badge for AI systems.

Modulus was described as the machine learning platform and Modulus Labs as the company developing it. Its core product was Remainder, a zero-knowledge prover purpose-built for AI workloads: an integrating decentralized application supplies data, Remainder performs cryptographic checks to construct proofs of correctness, and the aggregated proofs are submitted on-chain for verification, using Ethereum for proving. Cited application areas included trustless NFT appraisal, authenticated machine-generated art, tamper-proof on-chain finance and lending collateralization, private identity authentication, AI-driven game economies, recommendation and matching algorithms for web3, oracles for off-chain data, and AI-influenced governance. According to Tools for Humanity, Modulus produced the first verifiable AI games and authenticated AI art and was the first team to prove a multi-billion-parameter large language model in zero knowledge.

In December 2024, Modulus Labs joined Tools for Humanity, a contributor to World (formerly Worldcoin), where the team became TFH's Applied Research group focused on cryptography and machine learning.

Founding story

Accounts of the company's origins center on co-founder Daniel Shorr, who was drawn into crypto during the pandemic and, with collaborators, studied protocol whitepapers with a focus on zero-knowledge cryptography. An early experiment, Rocky.bot, was an AI program built to predict the price of ETH; it was initially self-funded by the founders and later sustained by donations of thousands of dollars from interested observers before the experiment ended. That work led the team toward software for verifying AI outputs and enabling smart contract developers to integrate AI, which became Modulus. Sources differ on details: one account states the company was founded in 2021 by Daniel Shorr, Nicholas Cosby and Ryan Cao and that it came to wider attention in 2022, while Tools for Humanity states Modulus Labs was founded in 2022 by Stanford graduates Daniel Shorr and Ryan Cao, motivated by the growth of generative AI and the need for accountability tools.

Business model

Modulus offered proving infrastructure that other protocols and applications integrated: a partner dApp or on-chain service supplied data or model execution to be proven, and Modulus's Remainder prover generated zero-knowledge proofs verified on-chain. Public sources do not describe pricing or contract terms.

Traction

Remainder V1 was released and used by partners including Upshot, Worldcoin, Ion Protocol, Polychain Monsters, Spectral Finance and AI Arena, with a planned Upshot NFT price feed combining AI models and on-chain verification. Deployments with Tools for Humanity reached thousands of Orbs and, per TFH, supported verifiable AI authentications for individuals globally.

Latest developments

On December 12, 2024, Tools for Humanity announced that Modulus Labs was joining TFH to form a new Applied Research team dedicated to cryptographic and machine learning work for World, following a collaboration that started in spring 2023.