We work at the frontier of verifiable data.
Every engagement is applied cryptographic research. The hard problems behind a zero-knowledge database are not solved once and shipped — they are solved again, specifically, for each institution’s data, regulator, and risk perimeter. These are the questions we push on.
Arithmetizing SQL
Expressing relational operators — joins, aggregates, range filters — as efficient PLONKish custom gates and lookup arguments, tuned to the data domains of a specific institution rather than a generic benchmark.
Proving at scale
Recursive proof composition and hardware acceleration (GPU, FPGA) to bring proving for large, realistic query workloads from minutes toward seconds — the difference between a feasible engagement and an impossible one.
Live data
Commitment cadence and update models for datasets that change continuously, so that a single new write does not require re-committing the entire dataset.
Composing guarantees
Combining zero-knowledge proofs with differential privacy and secure multi-party computation, so a released result protects both the underlying data and the individuals inside it.
Verifiable AI
Proofs of training-set membership and verifiable inference, as high-risk AI moves from principle toward enforceable accountability under emerging regulation.
Trust topology
How verifiability collapses chains of auditors, attestations, and vendor assurances into a single checkable artifact — and what that changes for institutional governance.
A young discipline, built on rigorous ground.
Verifiable databases rest on a decade of peer-reviewed cryptography — zero-knowledge proofs, KZG and IPA polynomial commitments, PLONK and PLONKish arithmetization, the Halo2 proving framework, recursive composition — hardened securing real value under genuinely adversarial conditions. We build on that foundation and extend it where regulated data demands more than the literature has yet answered.
That is the line we work on: the distance between what has been proven possible in principle and what a specific institution can run in production, under its regulator, against its data.
We publish our thinking in the open.
We do not keep the field behind a paywall or an NDA. Our conceptual work is public — so a technical buyer can evaluate the depth before a single conversation.
The field from first principles
Plain-English explainers — proofs, commitments, PLONKish custom gates, lookup arguments, and where zkDB sits among FHE, MPC, TEEs, and differential privacy.
Where it meets regulation and practice
Our perspective on the regulatory landscape, procurement realities, and what verifiable data changes for the institutions that adopt it first.
Bring us your hardest data trust problem.
Briefings are confidential, deeply technical conversations with our principals. We work with regulated enterprises, central institutions, and serious research programs.