Own Product · Open Standard
openworkid — professional identity that belongs to you, not the platform
Who owns your professional reputation? Today it lives on LinkedIn, Upwork, with your employer — on infrastructure you don't own and can't take with you. openworkid flips that: an open standard and a reference platform where the profile belongs to the person, is portable, and isn't tied to any platform. Conceived, built, and operated end to end. And in a world where AI agents act on behalf of people, the question only gets sharper: whose identity do they act on — a platform's, or yours?
What was built
Three layers, each usable on its own, all built on the same principle — the identity belongs to its owner:
The standard — openworkid.org. A JSON-LD schema for portable identity: ExperienceRecord, PeerVerification, HumanProof, PortableIdentity, Extensions. Schema.org Person type, server-side rendered, machine-readably indexable. Openly documented so the identity exists independently of any single platform — like SMTP, not a proprietary network. The standard first, the platform as one implementation of it.
The platform — upstand.work. Full data portability as a JSON export — a basic right, not a premium feature. Human-readable, stable URLs that belong to the owner. Passwordless magic-link auth, so no credential store and a smaller attack surface. SSR plus Schema.org for clean rich results.
The MCP server. Seven tools across three auth levels, so AI agents can discover and read a professional identity — with cleanly graduated permissions instead of all-or-nothing, controlled by the owner. Integration examples for Claude, the OpenAI Agents SDK, and REST. MCP from day one, not bolted on later.
The hard decisions
The central decision is ownership as architecture, not as a promise. Open standard before implementation, so the identity outlives the platform. Export, stable owned URLs, and an open schema are built in, not retrofitted — portability can't be added later. And AI-native from the first commit: the MCP server is part of the architecture for an agent-mediated world, not an add-on.
An open professional identity raises a second question immediately: if the profile is yours and you fill it in yourself — how does anyone know it's true, in a world where AI can fabricate any profile perfectly? The answer is a trust mechanism deliberately built to not work like an endorsement: real, named people who vouch with skin in the game. That produces a trust graph that can't be replicated synthetically — and turns a self-hosted résumé into an identity you can trust.
Why this is here
I bring the same approach embedded in your team: treat architecture as a decision, not a diagram; portability and open interfaces from the start instead of lock-in; design AI as infrastructure, not as a feature. openworkid is that standard, carried through to production on a system of my own.
Open standard · portable by design · MCP from day one
See it yourself
The same for your team? Let's talk about it.
Send an inquiry