An applied cryptography studio designing infrastructure for industries that cannot afford to get it wrong. Threshold is the runtime layer we've been preparing to build for years.
Inversed is an applied cryptography firm. We've spent the last several years building cryptographic infrastructure for the people who needed it most — Worldcoin on biometric verification, Scroll on zero-knowledge VM protocols, regulated finance teams on private computation, and the open-source ecosystem on primitives the next generation of the internet will be built on.
Our work has always sat at the same intersection: privacy-preserving computation and provable trust. Two halves of the same question — how do you let systems do useful work on data they shouldn't see, and how do you let people verify that the work was done correctly without taking anyone's word for it.
Threshold is that question applied to AI agents.
For two years we watched the agentic AI category build a category of failure modes underneath it. Prompt injection. Intention drift. Credentials sitting on developers' laptops. Audit logs written by the same statistical engine that took the action. Skills pulled from random GitHub repos.
We were uniquely positioned to see this because cryptography is the field where these problems already have answers. Signed provenance. Capability-based authorization. Sealed execution. Tamper-evident audit chains. The primitives that make distributed systems trustworthy at scale are the same primitives that make agentic systems safe to ship at scale.
So we built it.
Cryptography is the field where these problems already have answers.
Many with PhDs from UC Berkeley, the Weizmann Institute, and other leading research programs. We helped organize Eurocrypt 2025, the premier European cryptography conference, hosted in Madrid.
Over a decade in cryptography. Founded zkProof.org. PhD-track cryptographer. Combines technical depth with the conviction that cryptography is, as he puts it, "not just technical — it's ethical."
Applied cryptographer leading high-assurance systems across zero-knowledge proofs, multi-party computation, zkVMs, and differential privacy. The architect of Threshold's runtime.
Every system we build reflects a choice about what kind of digital world we want to live in. We choose privacy as a foundation, not a feature. Verifiability as a default, not an upgrade. Intention as a first principle, not an afterthought.
We do not ship things we cannot defend. Cryptographic engineering is unforgiving in a way most software isn't — a single broken primitive ripples through every system built on top of it. We work slowly enough to be right, and fast enough to matter.
The best cryptography is invisible. Users don't think about it. Builders don't have to reason about it. Threshold is designed to disappear into the agent runtime so people running AI-native companies can think about the company instead of the infrastructure.