The setup
The idea grew out of my ongoing study of economics and two frustrations I kept hitting. First, the inability to transfer purchased digital assets: a lifetime of digital purchases that, unlike physical property, can’t be capitalized, resold, or passed on as inheritance. Second, I watched friends trying to buy property overseas run into the wall of having no portable, verifiable digital identity. That was money spent that never became an asset you could hold, capitalize, or transfer. At the same time I was, and am, a firm believer in secure communication between trusted parties. I saw that pairing it with identity validated by attestation opened up far more than private messaging: third-party-mediated business communications, and above all revocation. If entitlements are tied to attested identity attributes such as employment, a degree, or an email address, they can be revoked the instant the attribute changes. That pain was everywhere. In the pre-federated-authentication era, companies routinely had former employees retain access to sensitive entitlements, licenses, and services long after termination. I believed a single unified system could tackle all of these at once, one that was identity-proofed, attestation-based, and revocable. So I built it, leaning on some of the KYC patterns I’d built at Zecco.
Triverium’s premise was identity-proofed secure communications: binding communications and digital assets to verified identity through an issuer-controlled, revocable attestation model. You could establish who someone was, and the issuer could revoke that trust after the fact. At its core it was the distribution of attestations through trusted intermediaries. That mechanism made communications truly secure, and it generalized cleanly to third-party-mediated business communications and to revocable entitlements. I designed and built it, and raised money for it, as founder and CTO.
The thesis was genuinely ahead of its time. What I specified in 2012, issuer-controlled, revocable digital credentials/attestation, is essentially the model the EU would mandate a decade later under eIDAS 2.0. That is EU Regulation 2024/1183: certified identity wallets with issuer-controlled revocable credentials, and public-sector acceptance from late 2026. I was right about the problem years before the market and the regulators arrived at it. The world ultimately standardized on phone-wallet formats, SD-JWT VC and ISO mDL, rather than the X.509 attribute-certificate approach Triverium used. The model was right even where the encoding the market chose differed.
My role
Founder and CTO. I designed the architecture, wrote the software, and did the fundraising. A classic one-founder-wearing-every-hat startup. Triverium’s own development and fundraising continued in parallel throughout the consulting period below, and the consulting paid the bills while the product kept moving.
What I built
A real .NET 4.x / WCF estate, ~21,750 LOC across six folders, not slideware:
- SecureSend, the secure-messaging product, including a polished Windows client with layered Application/Domain/Presentation, MVVM, and custom WPF controls.
- SOFT / FastSend, a high-performance, resumable, chunked large-file transfer engine. It was a staged pipeline with queue-depth backpressure through
QueuedReader/QueuedProcesser/QueuedTransmitter, independently addressable chunks carrying id + offset + length + digest per block, a bitmap “resume” sidecar with one bit per block, and pluggable compression. I designed it for 64-bit, exabyte-class scale and exercised it to ~107 GB transfers across 284 preserved benchmark runs in 2014. - CertServices, the attestation / certificate-authority layer underneath it all.
A nice engineering nugget from the SOFT benchmark data: parameter choice alone moved wall-clock throughput >10× on identical input. That proved the premise for a self-optimizing transfer, a direction the data justified, even though I never built out the auto-tuning feature itself.
Net result
Working software and a prescient thesis, but it didn’t reach escape velocity, and I wound it down as I closed it out and moved to RelayHealth, ~late 2014/early 2015. The deeper blocker wasn’t the technology; it was legal and economic. The economic premise around transferring/reselling digital goods ran into settled copyright law. Capitol Records v. ReDigi, SDNY 2013, aff’d 2nd Cir. 2018, held that digital resale requires making a reproduction and therefore falls outside first-sale doctrine, and that changing it is Congress’s job, not the courts’. Looking back: the binding constraint was never the cryptography, it was the law, and it still is.
It drew real interest. Through Shireen Piramoon, I got in front of Samsung Strategic Ventures at an SF startup event, and later connected with other investors; Samsung in particular was keen. Rebekah Wu of RH Partners organized an introduction to Ray Rothrock at Venrock, the veteran cybersecurity investor. The sticking point was hardware. Triverium’s security model leaned on an on-device HSM/TPM, and Samsung would not expand their device bill-of-materials to include one. I also landed a real adopter: my friend Tim Jorgenson got his company, VTrader, to use it. But funding ran short, and as it did, Craig Schultz and his team approached me for consulting; through those connections, more consulting opportunities followed. That is how Triverium quietly became the entity I consulted through for the 2012 to 2014 run of engagements.
The story / what I learned
- Being right early is not the same as being right on time. I specified issuer-controlled revocable attestation in 2012; a regulator effectively ratified the idea in 2024 with eIDAS 2.0. The thesis is the single appreciating asset, but the window, the encoding, and the delivery channel all moved.
- Know whether your constraint is technical or legal. Triverium’s was legal: first-sale doctrine for digital goods. No amount of cryptography moves a constraint that lives in statute and case law.
- A one-founder startup is a forge. Design, build, and fundraise at once. It’s where a lot of my range comes from, and it’s why “Founder/CTO” means hands-on here, not titular.
- A hardware dependency is an adoption tax. Triverium’s reliance on an on-device HSM/TPM was exactly what a strategic investor, Samsung, wouldn’t swallow; they wouldn’t change their device bill-of-materials for it. The market wanted a software path, which is the road the industry ultimately took with phone wallets and software credentials.
- The bitcoin one that got away. By the back half of Triverium I was mining BTC as a side project. When I brought Nicholas Menaker, my old Corrigo CTO, in to gut-check the company, one of his two pieces of advice was take $10K and buy bitcoin, then about $800 a coin. I laugh about it now.
Key people
-
Paul Vilevac, sole founder.
-
Simon Shenkymen, recruited to assist on the build.
-
A former mentee did admin work at Triverium while I taught him the trade, the start of a mentorship that ran on through WatchGuard to Nike CIS Ops.
-
Shira Abel, a friend whose then-new firm built Triverium’s brand book; Paul hired her early, when she was first getting her firm going. linkedin.com/in/shiraabel
-
Nicholas Menaker, Paul’s former CTO/boss at Corrigo; stepped in to help during a difficult stretch.
-
Samar Kawar, from Paul’s GoldMail days; also lent a hand.
-
Shireen Piramoon, from the GoldMail branding work, routed investor interest Paul’s way at an SF startup event, including Samsung Strategic Ventures.
-
Tim Jorgenson, friend and VTrader colleague; got his company to adopt Triverium.
-
Rebekah Wu of RH Partners, who organized the Venrock introduction. linkedin.com/in/wurebekah
-
Craig Schultz, in Paul’s orbit from RMS → NeoPrime; his team’s approach for consulting became Triverium’s pivot.
Technologies used
.NET Framework 4.x, WCF; WPF/MVVM desktop client; a staged, back-pressured, resumable chunked-transfer pipeline with per-block digest + bitmap resume; pluggable compression with LZ4/LZMA/Zip; a certificate-authority / attestation layer; issuer-controlled revocable digital-credential model.
Patents / awards
Several inventions were developed; none were filed.