Products
we build our own software.
su.engineering is not only a client-services studio.
We also design, build and operate our own products. This keeps us close to the same realities our clients face: architecture, security, product decisions, infrastructure, distribution, support and the tradeoffs required to turn an idea into software people can actually use.
Our products are also where we test many of the ideas that later inform client work.
Velum
LIVE
Privacy infrastructure for everyday AI workflows.
Velum helps prevent sensitive information from being exposed when people use AI tools and other external services.
It gives us direct experience with privacy-first architecture, local processing, desktop software, browser integrations, cross-platform engineering, enterprise security requirements and shipping software to real users.
What this proves
OpenKYC
APPROACHING LAUNCH
Reusable identity verification without repeating KYC.
OpenKYC is a credential-based identity and age-verification system designed around reusable proofs rather than repeatedly collecting identity documents and biometric data.
The product brings together digital identity, credential issuance, verification, privacy, wallet flows and standards-based interoperability.
OpenKYC is also a practical environment for testing the identity architectures and interoperability problems that appear in client work.
What this proves
REM
IN DEVELOPMENT
Shared engineering memory for teams working with AI agents.
REM explores how engineering knowledge can persist across developers, repositories, machines and AI coding tools while remaining trustworthy, reconciled and useful.
The work combines agent infrastructure, retrieval, knowledge reconciliation, developer tooling, privacy-aware processing and shared context systems.
What this proves
Klangfarbe
IN DEVELOPMENT
Verifiable identity for AI voice agents.
Klangfarbe is an open protocol for carrying a verifiable binding between an authorized AI agent and the speech it generates, so the receiver of a call can check which organization's agent is speaking and for what purpose. A working proof of concept is in build now.
The work joins verifiable credentials, workload identity, federation and trust frameworks with live audio binding that has to survive WebRTC, SIP and telephony codecs. A missing binding is reported as unverified, never as human.
What this proves
products are part of the engineering model
Building our own products changes how we approach client work.
We have to live with the consequences of architecture decisions, maintain software after launch, handle real users, deal with distribution and security, and make tradeoffs when time and resources are finite.
We bring that product discipline into project work.
Discuss a product or project.
Hard problem? Unclear architecture? Missing specialist capability?
hello@su.engineering