steelman each, then the wedge · plain-text master: battle-card.md · 2026-07-07
One page. Steelman first, then the wedge. Numbers sourced in brand/research/market-research.md.
The map in one line: the market splits into wrapper-minters (form the entity, then nothing), post-incorporation operators (assume the entity already exists), and agent plumbing (move an agent's money or label its traffic). Nobody does formation → immediate operations → optional incorporation. Nobody admits agents as registered actors.
Steelman: Atlas is the category king — 100k+ founders, Cooley-grade docs, +130% YoY growth Q1 2026, and distribution gravity into the Stripe stack. LegalZoom is a $756M/yr machine. Doola owns the non-US-founder wedge (175+ countries). Clerky owns VC-track correctness ($100B+ raised by its startups).
But structurally: all four are human-only (KYC on a natural person), effectively single-jurisdiction (Delaware/US), and nothing exists before the state filing. Their "operations" are bookkeeping and tax paperwork — none runs a live treasury, governance, membership, or fundraising. Atlas's own copy: "Bank, fundraise, and charge customers immediately after incorporating."
The wedge: we start before their product begins and keep running after it ends. "Incorporation is a feature, not the front door." Their growth (+130%) is our why-now evidence.
If they build it: starting before the filing reframes their own funnel — incorporation is the product they sell. And an agent-inclusive registry needs neutrality Stripe can't offer (see agent rails below).
Steelman: Carta is ~$442M ARR, 50k companies, the de-facto standard cap-table; AngelList administers ~$170B; Fairmint is an SEC-registered transfer agent pushing an open cap-table protocol.
But structurally: the same gap inverted — Carta Launch requires a certificate of incorporation just to onboard; AngelList RUVs only fund already-incorporated companies; Fairmint assumes a US corporation. All human-only, all US-securities-centric, all private proprietary ledgers. And Carta's Jan 2024 scandal — an employee pitching a customer's investors with confidential cap-table data, Carta exiting secondaries "to prioritize trust" — is the concrete argument for a public registry: the record, not the vendor, is the trust anchor.
The wedge: the whole venture stack starts after the moment Entity.ID starts. We are the record they all wish they could query.
Steelman: these are giants moving fast — all four payment stacks shipped agent rails within 18 months; Skyfire coined product-grade KYA; Catena raised $30M (May 2026) and is applying for an OCC trust charter; Microsoft/Okta ship agent IAM today.
But each proves exactly one narrow thing: AP2 mandates → a human authorized this purchase. Cloudflare signed agents → this request came from a known operator. KYA vendors → this developer passed review. Entra/Okta → this agent has a login, inside one tenant. Catena → this agent has an account — under Catena's charter, never its own standing. Nobody proves this agent is someone — with owners, a treasury, rules, and a jurisdiction.
The wedge: every rail presupposes a legally accountable party behind the agent that none of them provides. We're the presupposed layer — and the natural partner to all of them (registered ventures are what an agent-native bank or payment rail wants on the other end).
Why not them: neutrality. Visa won't gate on Mastercard's attestations; platform identity dies at the tenant edge. The standing layer structurally wants a neutral open standard — that's the Entity.ID protocol.
Steelman: MIDAO has a real government partnership (RMI) and reports rising AI-agent incorporations — direct demand validation. Otoco mints a wrapper in under a minute for $99–299/yr, and an agent can trigger it.
But: MIDAO is $9,500 + $2,000–5,000/yr, concierge-manual, one jurisdiction that tightened in 2025 — and mandates human kill switches: the agent is never the member or principal. Otoco's "agent support" means an agent can buy paperwork. Otonomos formed ~1,029 entities in a decade. Wrapper-only is a small business; nobody puts agents inside entities with roles, treasury mandates, and governance-scoped permissions.
The wedge: they sell the agent a document. We make the agent a registered actor.
Steelman: the honest default. Most of the 30.4M US zero-employee businesses never incorporate and feel fine; an LLC is a solved product at $0–500.
But: informal teams are involuntary general partnerships — unlimited personal liability (Samuels v. Lido DAO, 2025: DAO participants held liable as general partners), unwritten ownership, no treasury, no recourse. And an LLC is a certificate that gathers dust: no members/governance/treasury software, franchise obligations from day one, one jurisdiction. For agents, do-nothing isn't an option at all — no legal existence is the status quo.
The wedge: structure at the effort-level of a group chat, at the moment of the handshake — and the record grows into incorporation instead of replacing it.
"Formation platforms end at the filing. Equity tools can't start until the filing. Agent rails move money for actors that legally don't exist. Entity.ID is the registry where ventures and agents actually exist — operating from minute one, incorporating when ready, checkable by anyone before value moves."