Who may set the rules
Articles, bylaws, an operating agreement or a board action can recognize the company's Corporate Digital Name and authorize an authority policy. Counsel decides which instrument is right.
A key can sign. A tenant can log someone in. Neither says who authorized the act. Most governance sits beside the work — a policy document, a log someone reads afterwards. A log file is governance that stays home. CorpID puts the company's authority inside each signed act, so whoever receives it — a customer, auditor, regulator, or their AI — can check it before acting: proceed, refuse, or send for review.
The company's own corporate documents authorize a plain, technology-neutral process for acting digitally. The written authority policy can then change as tools change — Microsoft, Google, a chain, a credential format — without rewriting the company's founding documents. Every change to it is itself a signed act.
Articles, bylaws, an operating agreement or a board action can recognize the company's Corporate Digital Name and authorize an authority policy. Counsel decides which instrument is right.
Who may connect an identity provider? Who may authorize an AI agent? Which acts need a person to approve first? How are keys, tenants and people changed or removed?
Each signed authorization names the policy version that allowed it. Forty days later, whoever relies on it can check the authority in force at that moment — judged by the rules then, not today's settings.
The policy sets out who may do what, and when a person must approve. It is a set of decision rules, not a catalog of cryptographic switches.
| Policy area | Question it answers | Why it matters |
|---|---|---|
| Corporate Digital Name | What durable identifier does the corporation recognize? | All provider and actor bindings converge on one corporate principal. |
| Binding authority | Who may connect Microsoft, Google, AWS, Okta or another identity provider? | Stops an administrator from quietly becoming the corporation. |
| Delegation authority | Who may authorize people, workloads, machines and AI agents? | Turns "has a login" into authority with stated limits. |
| Escalation | Which acts need a verified person to approve before they happen? | A missing approval shows up instead of staying invisible. |
| Revocation | Who can stop authority immediately, and what record follows? | Authority must be removable without replacing the corporation. |
| Rotation | How do keys, providers, officers and agents change over time? | The corporate identity should survive all of them. |
| Continuity | What happens on merger, acquisition, dissolution or provider exit? | The record must say who succeeds, and when authority ends. |
This is a product template, not legal advice. The point is a policy a machine can read: signed, versioned, and findable by anyone who receives the company's work — without asking the company.
Every consequential authorization can name a policy hash and version, so a later reviewer does not have to guess governance from today's configuration. Each piece of evidence says what it proves and what it does not.
An employee can see whether their provider login only identifies them, or whether a signed authorization from the company lets them act for it in a specific class of action.
The agent carries a signed authorization from the company: its scope, conditions and expiry. The policy says who was allowed to sign that authorization. People won't check who authorized it. Their AI will.
Epistery already treats an identity as a contract with interchangeable device rivets. CorpID adds the corporation and a pointer to its policy; it does not replace Epistery's own authorization machinery.
Write the policy in Setup, sign the exact authorization, and let anyone check the result in the public verifier. Don't take our word for it — open the transaction.