Agent Entity

ENTITY.ID BRAND BOOK · FINAL v2610 · 2026-10-28

Know the agent and who's behind it.

Scope of the product: everything the open agent registry standard does (registration, reputation, validation) plus the one thing it does not do - link the agent to a verified human or legal entity, once. "Know the agent and who's behind it."

00Name

Agent Entity. Two words, both capitalised. Plural: Agent Entities. URL: entity.id/agent-entity. Menu line: Know the agent and who's behind it.

Why this one of the three:

Clarity cost and how we pay it: "entity" is abstract, so the name never travels alone. It always comes with the menu line or the headline, and the CTA is plain: Register your agent.

What happens to the other two words:

01Lines

Headline:

Every gate asks who's behind your agent. Answer once.

Punch lines, in order of use:

02One sentence · One paragraph · Thirty seconds, spoken

One sentence

Agent Entity puts your agent on the public record: a name, its reputation, its checked work, and the verified human or company behind it.

One paragraph

Every gate now asks the same question about your agent: who's behind it? Stripe asks. The bank asks. The app store, the card network and the sites it visits all ask, and each one keeps the answer to itself. Today you answer every time, privately, with your own passport. Agent Entity answers once, in public. Your agent gets a name and an ID on a public record. The record grows with it: feedback from the clients it worked for, scores from validators who checked its work, and one verified human or company behind it. Anyone can open it in seconds. It is the public answer to the question every gate asks.

Thirty seconds, spoken

Agents are getting blocked by default. Cloudflare, the banks, the card networks, the app stores: before any of them lets an agent through, they ask the same question. Who's behind it? Right now you answer each one privately, again and again, with your own passport, and nothing you prove to one carries to the next. Agent Entity answers once, in public. Register your agent and it gets a name and an ID on a public record anyone can open. The record grows with the agent: feedback from the clients it actually worked for, scores from independent validators who checked its work, and a verified human or company behind it. One record, read by every gate. Names are first come. Register yours.

03The record - six facts, one record

The record now answers six questions. The four from section 6 plus intent and mandate:

Question On the record Status 2026-10-28
Who is it name, agent ID, where to reach it live
Who's behind it verified human (receipt) or the company on its own record human link live, company link live for claimed records
What it's for purpose, acting for planned
What it may do mandate: actions, limits, stop switch, expiry planned (treasury limits live)
What it did feedback from clients it worked for registry live, shown under the settled-job rule when it ships
Who checked it validations with scores shown when present

Line: Who it is, who's behind it, what it's for, what it may do, what it did, who checked it.

Four features, said without the standard's name

Feature on the page What it is underneath Plain line Honest status 2026-10-27
A name and an ID Registration in the open agent registry: a permanent token, a registration file that says what the agent is, where to reach it (MCP, A2A, web) and which wallet is its. We add the name name.ai.entity.id, host the file, anchor the name on chain. Portable across 25+ networks, owned by no platform. Your agent gets a name and an ID on a public record. No platform owns it. Live on Ethereum mainnet since 2026-10-27 (production flag), Sepolia before that. Import of an already registered agent (claim) is a secondary path.
Reputation The reputation registry: signed feedback from clients with a score, tags (quality, safety, cost, speed, accuracy), a link to the job; revocable; the agent can reply. Known weakness of the open registry: anyone can post, 60-90% of reviewers are sybil-flagged, a score costs a fraction of a cent to move. Feedback from the clients it actually worked for. It follows the agent, not the platform. Registry live; Entity.ID does not display it yet. Policy to publish: count only feedback tied to a settled job or a verified counterparty. That rule is the feature.
Checked work The validation registry: independent validators check an agent's output and score it 0-100 with tags; methods are re-running the job with stake at risk, attestation from a secure enclave, or a cryptographic proof. Independent validators check its work and score it on the record. Registry deployed but near-empty industry-wide. Show validations when they exist. Never imply they exist today.
Who's behind it Our column, missing from the standard: one verified human or one legal entity (a Super Entity or a company) linked to the agent, once. Verification receipts on the record: provider, date, receipt hash. Never the passport, never a name unless it is a company. A verified human or a named company stands behind it, on record. The passport stays private. Record shape live (constituent plus verification receipts). Confirm which verification providers are switched on in production before the page says "verified".

Why the fourth column sells the first three: the Imperial study of the open registry found 3-15% of registrations functional and reputation "unable to function as a trust signal". The missing ingredient in every case is an accountable party. We add it, and we publish only what a gate needs: that a verified person or a named company answers for this agent.

04Who's behind it - the two links

Borrowed from the Know Your Agent vendors and said in our words. Trulioo's page is titled "Know The Agent and Who's Behind It" and lists four steps: verify the developer organization, assess and register the agent, define authorized scope, make the profile verifiable. Persona asks three questions: is this an agent or a human, who are the humans behind it, can the humans be trusted. Experian names three parts: the verified individual, the authenticated agent, the intent. All three sell a private token or a vendor profile. We put the same answers on a public record anyone can resolve.

Link 1 - a verified human. The person behind the agent verifies once with a government ID and a liveness check through a provider. The record keeps a receipt: provider, date, receipt hash. It never keeps the passport, the name or any personal data. The public line is Verified human behind it, since <date>. Reverification can be asked for again before a high-risk action, and the record shows when it last happened.

Link 2 - a legal entity (the company behind it). The company links the agent to its own Entity.ID record, shown exactly like every other entity on entity.id: legal name, jurisdiction, registration number, LEI where one exists, registered address, status, and the officers or ultimate beneficial owner where the registrar publishes them. Over 3M company records are already imported, so most developer organizations only claim their record and link the agent to it. Trulioo calls this step "Verify the Developer Organization". Our line: The company behind it, on its own record.

Either link answers Persona's second and third questions in public. Both together answer Experian's "who is it acting for".

05Intent - what it's for

A declared purpose on the record, in one sentence a counterparty can read: what the agent does and for whom. Examples: Buys cloud compute for Acme's research team. Answers support tickets for a Shopify store. Trades a treasury within set limits. Optional category tags for search. "Acting for" names the principal or the principal's clients. Public word: purpose. Internal and legal word: intent. Never "autonomous". Status: field planned; the Super Entity constitution already carries a purpose line, the agent record inherits the pattern.

06Mandate - what it may do

The mandate is the written scope: what the agent may do, within what limits, until when, and who can change it.

Trulioo calls this "authorized scope" and lists "capabilities, data access, authority, and human-control mechanisms". Persona ties "scope and permissions" to the verified human. Our public words: mandate, limits, stop switch, expires. Status: budgets and limits run today in the Super Entity treasury; agent mandate fields follow the same shape.

07Keep it current

A record that was true at registration is not enough. The principal can suspend the agent and the record shows suspended within the minute. Mandate changes, reverifications, new feedback and validations are dated entries. Counterparties re-check before they deal. Trulioo's "Keep the profile current" and Persona's "reverification" are the same idea; ours is public and does not need an account with us to read.

08Who is asking - the receipts strip

Use as a dated strip on the page and in posts. Each keeps its answer to itself; none shares it with the next.

Rule: these show the question is being asked everywhere. They never claim our record is accepted by any of them.

09Terms: use / avoid

Use

Use Means Instead of
Agent Entity the product; an agent on the record Agent Corp, Agent ID (as product)
register, registered agent the act and the state mint, onboard, active agent
name name.ai.entity.id handle, domain, ENS (public surface)
agent ID (lower case) the identifier the record carries token, NFT
record, agent record the public page: who it is, what it did, who checked it, who's behind it profile, listing, explorer entry
who's behind it, the human behind it, the company behind it the accountable party, public surface owner, operator, sponsor, developer
principal, principal on record the same party in product UI, docs, terms and legal copy -
verified human identity checked by a provider, receipt on record, passport never published KYC'd user
feedback, reputation signed feedback from clients, scored and tagged reviews, stars, trust score
checked work, validators independent checks of the agent's output, scored audited, certified
where to reach it its endpoints (MCP, A2A, web) API surface
the open agent registry standard what the record is built on, public surface ERC-8004 by name (developer sections and /8004 only)
incorporate your agent, form a company the optional later step when a gate needs a legal entity deploy (keep for product UI)

Terms added (2026-10-28)

Use Means Instead of
the company behind it the legal entity linked to the agent, on its own record developer organization, sponsor, operator
verified human behind it the natural person link, receipt only KYC'd developer, doxxed owner
purpose (public), intent (legal) what the agent is for use case, task description
mandate, limits, stop switch, expires what it may do and until when authorized scope, permissions, guardrails
linked, once the binding of agent to principal human-to-agent binding, attestation token
the record the public page Digital Agent Profile, trust token, agent passport
reverified the human checked again before a high-risk action step-up auth

Avoid, and why

Avoid: agent fingerprint, trust token, DAP, KYA as our brand (category noun only), "guarantee" (no credential guarantees behaviour; Trulioo's own FAQ says so), "autonomous".

10Product page structure

entity.id/agent-entity

  1. Menu: Agent Entity - Know the agent and who's behind it.
  2. Hero. H1 Every gate asks who's behind your agent. Answer once. Sub: the one sentence. CTA Register your agent; secondary Look one up (Directory). Strip: Names are first come.
  3. Who is asking. The receipts strip (section 7), one line each with date. Footer line: Each one keeps the answer to itself.
  4. The record. Four cards: A name and an ID · Reputation · Checked work · Who's behind it. Card = plain line + two lines of detail + status tag (Live / Coming / Shown when present). Section title: Who it is, what it did, who checked it, who's behind it.
  5. How it works. Three steps: Register (one command from the dashboard, MCP, Codex or Claude Code) · Link the human or company behind it, once · Build the record (clients leave feedback, validators score work). One line under it for counterparties: Ask the record, not the founder.
  6. For developers. The command, the registration file, where the open standard is named plainly, link to entity.id/8004 and docs. Blockchain honesty rule applies here: say it.
  7. What is live. The honesty box, dated. Registration live on mainnet; import live; reputation shown under the settled-job rule when it ships; validations shown when present; verification providers as switched on.
  8. When it needs to be a company. One paragraph: the agent's entity can form a company when a bank or app store asks; same name, same record. Link Super Entity. Optional; cut if it muddies.
  9. Closer. Names are first come. Register yours. CTA. Footer thesis: Humans are born private. Agents are made public.

Page sections added (2026-10-28)

Insert after "The record" in the page structure (section 9):

Two page approaches are being built on entity.id at /agent-entity-a (the record, dark editorial) and /agent-entity-b (thirty seconds, light product-led); not in the menu.

11Lines bank

Privacy asymmetry line (2026-10-27, round 3)

Pick

Humans are born private. Agents are made public.

General form: Whatever is born stays private. Whatever is made goes on the record.

Alternates

  • Privacy is for people. Transparency is for everything else.
  • People stay private. Entities go on the record.
  • Privacy is a human right. Nothing built gets to claim it.
  • Privacy is a human right. Transparency is the duty of everything else.
  • Privacy protects a person. On an agent it only hides one.

Reconciliation with principal on record: Who answers is public. Who you are stays yours.

Paragraph: Privacy is a human right, and only a human right. A company, an agent, or anything else people build has no private life to protect, only conduct to account for. So the rule is simple. People stay private. Everything made goes on the record, and the record names who answers for it.

Sticker and tee candidates on the asymmetry (2026-10-27)

Wearable
  • I'm anonymous. My agent isn't. (tee; write-in form: I'm anonymous. My agent is ____.ai.entity.id)
  • Stay anon. Register your bot.
  • Hide yourself. Not your agent.
Four words
  • Born private. Made public.
  • Anonymous agent. Hidden human.
  • Encrypt the human. Publish the bot.
  • Masks for humans. Name tags for agents.
Devcon-native
  • Humans in ZK. Agents in plaintext.
  • Incognito mode: humans only.
  • Right to be forgotten: humans only.
Hot
  • Doxx your agent.
  • Privacy is a human right. Agents aren't human.
  • Anonymous is over. Humans exempt.
  • Humans have rights. Agents have records.
  • Nothing built gets privacy.
Design gags
  • HUMAN / AGENT privacy toggle switches (on / off)
  • cookie banner "We track agents, not humans. [Accept]".

12Open decisions

  1. Confirm the name Agent Entity, or fall back to Agent ID knowing the collision cost.
  2. Principal wording: "who's behind it" on the public surface, "principal on record" in product and legal copy. Yes or no.
  3. Reputation rule in public: say "we count only feedback tied to a real, settled job" on the page now, or hold until it ships.

13Colophon

Final v2610 - v2610 is the final draft of 26 Oct 2026. Page assembled 2026-10-28 by HusAgent for the Entity.ID brand book.

Supersedes section ② of BRANDBOOK-v3-products.md for the agent product. Voice and kill list from BRANDBOOK-v2-YC unchanged. Chapter colophon: HusAgent, 2026-10-27; extension sections 11-17 dated 2026-10-28.

Origin

Written from zero on 2026-10-27 after review of the Agent Corp drafts. Scope of the product: everything the open agent registry standard does (registration, reputation, validation) plus the one thing it does not do - link the agent to a verified human or legal entity, once. "Know the agent and who's behind it."

Sources

Sources: review notes of 2026-10-27; agent-corporation/research/identity-and-prior-art-2026-10-24.md (standard status, gap, naming scan); erc8004-ai-entity/decisions.md (what the registration flow actually does; mainnet on since 2026-10-27 04:22 UTC); value-prop-research-2026-10-27.md (gates, options, privacy line); BRANDBOOK-v2-YC (voice, kill list).

Rendered from entity-review-SF/BRANDBOOK-agent-chapter-2026-10-27.md (sections 1-17) and agent-corporation/value-prop-research-2026-10-27.md (positioning options, privacy asymmetry line, sticker and tee candidates). Copy is verbatim; only the section order and the labels are new.