The graph must deserve the trust it holds.
A professional graph can become abusive — hidden blacklists, opaque scores, surveillance. GRESS holds sensitive professional continuity, so privacy, provenance, and consent are architecture, not policy pages.
The core principle
Worker-owned.
Institution-verified.
Workers should not depend forever on an old employer's HR portal to prove their own history. But workers should not be able to fabricate institutional facts either. GRESS mediates both: the authoritative issuer attests, the worker receives and controls the record, third parties verify its provenance, and issuers can correct — with a visible audit trail.
Your history is yours
Meaningful history is exportable and portable. GRESS does not build its moat on imprisoning user data — open portability, with the intelligence and workflows as the reason to stay.
Sharing is a decision
You know what is shared, with whom, for what purpose, and for how long. Minimum-necessary disclosure by default — never a buried checkbox.
History is never rewritten
Corrections and revocations append; they don't erase. Current state reflects the latest valid condition, and material events stay tamper-evident.
Provenance tiers
Not all facts are equal — and we never pretend they are.
Every assertion in the graph carries its provenance, and the tiers are never visually collapsed. “User says they led a $5M project” is not equivalent to “employer attested it.”
Self-asserted
The person says it. Useful context — displayed as exactly that, and nothing more.
Counterparty confirmed
Another verified party affirms it — a manager, a client, a collaborator who was there.
Institution verified
An employer, school, certification body, or authoritative system attests the fact.
Transaction verified
A real economic or legally significant event occurred — through GRESS or a trusted integration.
Outcome verified
A defined outcome was delivered and accepted by an authoritative counterparty. The strongest evidence there is.
“Claims, evidence, verification, inference — always separate. AI inference is never presented as verified fact.”
From the GRESS data constitution
The data constitution
What we commit to.
- Provenance always — self-asserted and verified facts are visually distinct
- User permission controls sharing; users can export meaningful history
- Users can challenge incorrect assertions; issuer corrections are auditable
- Credentials can expire, revoke, and correct — with the trail preserved
- Interpretable trust signals — identity confidence, repeat relationships, dispute state
The refusals
What we will never build.
- Hidden employer blacklists or anonymous negative professional scores
- A universal social-credit-style score compressing a person into one number
- Inferred medical, family, or political characteristics
- Undisclosed salary sharing or opaque hireability scoring
- Surveillance of anyone's life outside work
Where AI fits
AI on top of truth —
never instead of it.
AI in GRESS normalizes events, summarizes verified history, spots missing verification, and recommends transitions — grounded in observed professional state changes, not generic career advice. It sees only the context each task requires: a career recommendation doesn't need payment records, and a contract assistant doesn't need your alumni relationships. And its inferences are always labeled as inferences.
Hold us to it
Principles are only real
when they're tested.
Questions about how GRESS handles data, consent, or verification — or something we should refuse to build that isn't on the list yet? Tell us.