Skip to content
AGI Future Foundation PBC

Features · AGI Future Foundation

The Mission Gate — The Seven-Stage Flow That Decides What Gets Built

Every company the Foundation's venture studio might ever build has to pass through the same narrow door, and the door is not a formality. It is a seven-stage flow — intake, mission screen, legal assignment, series designation, Steward Council review, human decision, execution and record — and at its centre sits a single human judgment: does this project advance safe, equitable, and socially beneficial artificial intelligence? If the answer is no, the project stops. It does not get a series number, it does not get a bank account, and it does not get built under the family name. That is the mission gate, and this feature is about how it is supposed to work — and about the fact that, as of today, nothing has ever passed through it.

AI-generated video — illustration, not footage of a real factory, event, or dashboard.

The door that only opens one way

Most venture studios start with the same question a founder asks in a pitch meeting: can this make money. The Foundation's studio starts somewhere else, and the difference is not decorative. Before any project can become a series, before it can take investment, before it can hire anyone or sign anything, it has to survive a review whose only question is whether it serves the mission — safe, equitable, and socially beneficial artificial intelligence, held to standards the Foundation sets. Money is not the question at that stage. Market size is not the question. The question is fit, and fit is decided by humans on the Foundation board, not by agents, not by a vote, and not by momentum.

The design is deliberately old-fashioned in one respect and deliberately strange in another. The old-fashioned part is the gate itself: a board, a minute book, a recorded decision with reasons, the way charities and companies have made consequential decisions for a century. The strange part is what feeds into it. Agents do the drafting, the research, the synthesis, and the preparation at every stage. A proposal may be written entirely by agents, researched by agents, and routed by agents — and still mean nothing until a human adopts it, a human screens it, and a human signs the decision. The gate is the point where the machine stops and the people start, and the design says that point is not negotiable.

That matters because the studio is designed to move fast. Agent-managed work is how the model scales: many series, few humans, agents doing the drafting, building, monitoring, and reporting across all of them at once. Speed like that needs a brake that cannot be talked past, and the mission gate is the brake. It sits above the six stage gates of the studio — Idea, Validate, Build, Launch, Operate and Scale, Exit or Wind-down — and it sits before them too. A project meets the mission gate at intake, before it is a series, and it meets it again whenever the Foundation board chooses to withdraw approval. A project that loses mission approval loses Foundation support and Foundation branding, even if AGI Corp wants to continue it commercially. The commercial head cannot overrule the mission steward. That is the whole design in one sentence.

There is a reason the gate is drawn this way, and it has to do with what the Foundation is. As filed, it is a California Nonprofit Public Benefit Corporation, and its board owes fiduciary duties to the charitable mission and the public, not to any other entity's shareholders. If the Foundation could be overridden by a commercial head, or if its mission screen were a rubber stamp, its directors could be found to have breached those duties. The gate is therefore not just good practice. It is the mechanism by which the Foundation stays a steward instead of becoming a captive. Counsel still has to design the control mechanism that makes that work in law — a mission-lock agreement, a veto over mission changes, independent directors — and that design work has not been done. But the gate, as a piece of paper, is already drawn.

So this feature follows a proposal through all seven stages, in order, the way the intake document and the operating model describe them. It also says, at every stage, what does not exist yet. That second thread is not a footnote. It is the point. A process nobody has used is a hypothesis, and the honest way to write about a hypothesis is to describe it precisely and then say, plainly, that it is untested.

Stages one and two: intake, then the mission screen

Stage one is intake, and intake is paperwork — deliberately so. A proposal is filed with five facts: the project name, a description, the founder or lead, the intended world or product, and the funding source, with one primary source named. The funding source has four possible values in the intake checklist: donation, grant, investment, or revenue. One must be named. A proposal that names no primary funding source is incomplete at intake and goes back for completion rather than forward to review, because the funding source decides which gate the project will later be routed through — the Foundation gate for donations and grants, the AGI Corp gate for investment and SAFEs — and a project that cannot say where its money would come from cannot be routed at all.

Agents do the intake work. Research synthesis scans and drafts. Proposal drafting writes the proposal in its defined shape. Task routing makes sure it reaches the right reviewers. None of that is approval. Intake opens a register row, and the register row is a tracking row, not a filing. A series exists only when its master LLC's operating agreement designates it, its books are segregated, and, where the statute requires it, its designation is filed or recorded. Filing a proposal makes it a candidate. Nothing more.

Stage two is where the gate actually closes or opens: the mission-fit review. The Foundation board asks whether the project advances safe, equitable, and socially beneficial AI under the Foundation's mission and standards, and records the answer as approved, declined, or approved with conditions, minuted with reasons. The review is done by the Foundation board, not by agents and not by the studio. Agents may prepare the checklist and the supporting material, but they do not pass or fail a project on mission grounds. That distinction is the load-bearing wall of the whole design: the people who prepare the evidence are not the people who judge it.

A declined project does not get a series number. The intake document states that without exception, even for a commercially promising project. Mission fit is a gate, not a preference, and not a score that a strong market case can outweigh. Conditions, when they are attached to an approval, are written down with the approval, so later stages can check whether they were met. A project whose conditions are never met returns to the Foundation for a fresh decision rather than drifting forward on the strength of an old approval. The gate is not a single moment. It is a standing condition that has to stay true.

Stage two is also where the Fellowship boundary is enforced. The Fellowship is a separate church, not a series, and it is not put in the series registers. It has its own governance records, and any partnership or support runs under written agreements with an independence clause, not through intake. That matters for the gate because it keeps the mission screen honest: the Foundation cannot smuggle a church programme into the studio portfolio by calling it a series, and the studio cannot claim a church as one of its companies. The registers stay clean because the boundary is enforced at the door.

Stages three and four: one master, one series, on paper

Stage three is legal assignment, and its rule fits in one line: one project, one master, one series, minuted. Once a project passes mission fit, it is assigned to one master LLC and only one. Existing family series sit under the intended Wyoming W3 LLC. All new projects go into the intended Delaware cohort series company, under the naming convention C1S1, C1S2, C1S3, where C1S1 means Cohort 1, Series 1. The assignment is minuted with the reason recorded — for new projects, the standing direction that all new projects go to Delaware, unless counsel advises otherwise for a specific project.

The reason for the one-master rule is that Wyoming and Delaware series are different statutes, different master LLCs, different registered agents, and different books, and the two states' series do not automatically recognise each other's liability shields. Counsel must confirm cross-state treatment before anyone assumes a shield travels. A project assigned to both masters would have two books, two managers, and two answers to the question of who is liable — which is another way of saying it would have no answer at all. So the assignment is minuted, and the minute is the evidence that the choice was made deliberately rather than by drift.

There is a second trap at this stage, and the intake document names it plainly: the cohort label is not a legal name. C1S1 is a reference label. The legal name is the master LLC's name plus the series designation in its operating agreement — for example, a Delaware master LLC name followed by Series C1S1. Contracts, bank accounts, and SAFEs are never signed, opened, or issued in the name C1S1 alone, or in a project nickname alone. The Delaware master LLC's name is still to be chosen and its registered agent still to be named. Until those exist, no Delaware designation can be made at all. That is not a detail. It is a hard stop on the whole flow for new projects.

Stage four is series designation, where the project stops being a proposal and starts being a series — on paper. The master LLC's operating agreement is amended or appended to designate the series, with its purpose, assets, and liabilities segregated on paper. A DAO charter is written for that series: its name and nickname, its purpose, its master LLC, its agent team with roles and limits, its proposal process, its Steward Council and its two-of-three rule, its funds rules, its reporting duties, and its exit and wind-down terms, agreed in writing before launch rather than after a dispute. The legal manager or signer is named — a human, or AGI Corp under a written management agreement — authorised to sign contracts and bank documents for that series.

Designation also settles two things that are easy to leave vague and expensive to leave vague: money and intellectual property. Segregated books and a segregated bank account are opened for the series, only after designation, and no series funds go into the master LLC's general account, AGI Corp's account, or the Foundation's account. Intellectual property is settled on day one: who owns the project's IP — the series, AGI Corp, the Foundation, or a contributor — by signed assignment or license, not by assertion. Until the Council is named, consenting, and holding keys, the series is Proposed, not Operating, and it is capped at Gate 1: it cannot pass to Build as an operating company. Until a spending threshold is set, all spending requires Council approval, and the default agent spending limit is zero dollars without Council approval.

Stages five and six: the Council reviews, the humans decide

Stage five is Steward Council review, and it is the first stage where the people deciding are the people who will live with the consequences. With the series designated and the Council seated, proposals go to the Steward Council: three humans deciding by two-of-three signed votes on budget, charter, funds, major contracts, and pivot-or-kill recommendations. The review is the Council's own work. Agents prepare the proposal, the evidence, and the options, and the Council judges them. A DAO vote may run alongside as an advisory signal, but it is not the decision. A DAO vote does not sign a contract, move funds, or issue securities. Only the series' legal manager or Steward Council, with signed authority, does that. A token vote, a poll, or an agent proposal is not a corporate act.

The review also covers the funding gate, and confusing the two funding gates is, in the operating model's words, the fastest way to break the law in this structure. Donation and grant funding runs through the Foundation gate. Investment and SAFE funding runs through the AGI Corp gate. Never the other way round, and never from an unformed series. The Council checks that routing as part of its review, because a series that takes investment before it is formed, or takes donations as if they were equity, has already broken the structure it was designated under. The review ends when the Council is ready to decide. The decision is the next stage, not part of the review. Review prepares a decision. It does not make one.

Stage six is the human decision, and it is made by named humans, in writing, with signatures and a date, minuted with reasons. For a series proposal, the Steward Council decides by two-of-three signed votes. For portfolio matters — allocating studio resources, or a final kill or pivot where the series agreements give the studio that right — the AGI Corp board decides. For mission matters, the Foundation board decides, and it can withdraw mission approval; that project then loses Foundation support and branding, even if AGI Corp wants to continue it commercially. For partnership or doctrinal matters, the Fellowship board decides for its own programmes. No single human, and no agent, holds all four powers. That separation is the point of the studio model as a governed venture studio, rather than a single-person project list.

The available decisions are always the same: approve, approve with conditions, amend, decline, pivot, or kill. A decision to kill is recorded with reasons, and the model treats killing a project as a studio success, not a failure to hide. That line is worth pausing on, because it is the design's answer to the most common failure mode of a venture studio: the project that should have died in month three and is still consuming money in month eighteen because nobody wanted to be the person who killed it. Here, the kill is a minuted decision with reasons, and the reasons are the record that the system worked. A studio that never kills anything is not a studio that picks well. It is a studio that cannot decide.

The decision also names what happens next: who executes, by when, and what evidence must be filed. A decision that names no executor and no deadline is incomplete, and the record should say so. That rule sounds bureaucratic, and it is — deliberately. A decision without an owner and a date is a wish, and wishes do not survive contact with a calendar. The flow is built to produce decisions that can be checked later, by someone who was not in the room, against a record that says who did what and when.

Stage seven: only the decided step, then the receipt

Stage seven is execution and record, and execution is deliberately narrow. Agents carry out the decided step, and only that step, in the form it was decided — operating inside an approved budget, running an approved launch checklist, executing the amended version of an amended decision. Anything needing a signature, a bank transfer, a filing, a hire, a securities offer, or a doctrinal act stops and goes back to a human, because those acts are human-only by definition. An agent that finds the decided step cannot be executed as written does not improvise a substitute. It reports back, and the flow returns to the Council for a fresh decision. Improvisation is the one thing the flow forbids outright.

Then comes the record. Every proposal leaves a dated receipt: who drafted it, which agent and which human, what was decided, who decided it and when, what was executed, and what remains open. The portfolio dashboard is the standing form of that record — one row per series, with gate, lead human, lead agents, actual funds in and out, actual revenue or the words none yet, actual mission outcome, risks and blockers, and the next gate or decision. It is updated monthly, reports actuals only, never lists an unsigned partner, customer, or investor, and never carries the Fellowship as a portfolio row. The loop then repeats. A new proposal enters at stage one, and the record at stage seven is what proves the last one worked.

The receipt matters more than it looks, because it is the only part of the system that an outsider can check. A board minute can be read by a lawyer. A dashboard row can be read by an investor. A dated receipt can be read by a regulator. The design's honesty depends on those artefacts existing and being accurate, and the operating model is blunt about what counts as evidence: a gate is passed only when the human approver signs and the evidence is filed. An agent's readiness flag is not a gate pass. An agent's submitted flag is not proof. A filing is not done until proof — a confirmation, a receipt, an acknowledgement, a bank statement — is filed in that entity's pack.

There is a final discipline inside stage seven that ties the whole flow together. The dashboard never carries a projection in a revenue or funds-in column. It never lists a partner, customer, or investor who has not signed. A series with no Steward Council is capped at Gate 1. The Fellowship is not a portfolio row. Those four rules are small, and they are the difference between a record and a pitch deck. A pitch deck is allowed to show the future. A dashboard is not. The mission gate exists so that the future the studio builds is the one the mission allows — and the record exists so that anyone can check whether that is what happened.

Put the seven stages together and the flow is a loop with teeth at both ends. At the front, the mission screen can kill a project before it exists. At the back, the receipt can prove — or fail to prove — that a decision was executed as decided. In between, every stage has a named human, a signed vote, and a piece of paper. That is the design. What it is not, yet, is a record of anything actually happening.

Why a decline is the system working

There is a temptation, when writing about a gate, to treat every decline as a failure — a good idea turned away, a founder disappointed, a door closed. The design treats it the other way round. A gate that never declines anything is not a gate. It is a corridor with a sign on it. The mission screen exists precisely to say no to projects that would be profitable, popular, or technically impressive but do not advance safe, equitable, and socially beneficial AI. If it never says no, the mission is not being enforced. It is being advertised.

That logic extends to the kill decision at stage six and to the Foundation's power to withdraw mission approval. A project that loses mission approval loses Foundation support and branding even if AGI Corp wants to continue it commercially. That is a deliberately uncomfortable power, and it is uncomfortable on purpose: it is the only thing that makes the Foundation a steward rather than a logo. A steward who cannot withdraw support is not a steward. A mission gate that cannot close is not a gate. The design knows this, and it writes the discomfort into the documents rather than smoothing it over.

The same logic applies to the two funding gates inside the flow. Donations and grants may fund a studio project only as a grant serving a charitable purpose, with reporting — not as studio equity, not as a SAFE, not as a subsidy for a commercial series' losses, unless counsel advises a specific lawful structure in writing. No SAFE, equity, profit share, or return may be issued by the filed nonprofit Foundation or by the Fellowship church, and no tax-deductible donation may be claimed, because no IRS 501(c)(3) determination letter is in the files for either. Those are not footnotes to the funding story. They are the funding story, for the nonprofit lane, and the mission gate is what enforces them: a project that cannot name a lawful funding source at intake cannot be routed to either gate, and a project that cannot be routed does not proceed.

It is worth saying plainly what the gate does not do. It does not judge whether a project is a good business. It does not judge whether a project is technically clever. It does not judge whether a project would make the family famous. It judges one thing — mission fit — and it judges it by a standard the Foundation board sets and is accountable for. That narrowness is a feature. A gate that tried to judge everything would judge nothing well, and a board that had to be right about markets, technology, and mission all at once would be right about none of them. The mission gate stays in its lane so that the other gates — the stage gates, the funding gates, the Council's budget votes — can stay in theirs.

And that is why the honest ending of this feature is not a success story. It is a list of things that have not happened. No series has been designated under this intake process. No proposal has moved through it. No Steward Council has been seated. No DAO charter has been signed under it. No money has moved under it. The seven stages are drawn, the checklists are written, and the registers are templates with blank rows. The first proposal to move through the flow will be the first test of whether the gate works — and the record at stage seven is what will prove it, not this page.

What does not exist yet, and what has to happen first

Because this feature describes a process, it is worth listing what the process has not yet produced, in one place, without softening. There is no designated series under this intake process. The Wyoming W3 register and the Delaware cohort register are templates with blank rows, and the April 2026 internal list of 33 named W3 sub-series is an internal printout, not a Wyoming filing and not a designation. Other materials have mentioned 35 cells and 34 worlds; the counts do not agree, and this feature follows the internal list of 33 named sub-series. There is no Delaware master LLC name, no Delaware registered agent, and no Delaware operating agreement on file. There is no Wyoming W3 operating agreement on file.

No Steward Council has been seated for any series, no DAO charter has been signed under this process, and no legal manager has been named for any series. No segregated series bank account has been opened under the studio model, and no money has moved under it. No SAFE, equity, or grant has been issued or made under it. No project has passed any stage gate under it, because no project has entered it. The valuations and raises in the corporate paperwork are planning figures, labelled as suggested, with no revenue, appraisal, or signed terms behind them. None of this means the process is wrong. It means the process is untested, and presenting it as tested would be the error.

What has to happen first is the order of work the paperwork itself sets out. Form the entities first, with counsel choosing one corporate form for each, and only then pursue the raises, the sites, and the build-out in the order the gates require. Counsel must design the nonprofit-control mechanism, or the head must be restructured into agreements between equals. Counsel must choose one PBC track in writing before any solicitation, and reconcile the site copy with the choice. Counsel must confirm the Wyoming and Delaware requirements for each master before any designation is treated as complete. The Foundation's good standing, Statement of Information, Attorney General registration, and IRS posture must be confirmed before any donation round or bank application.

There is also a question the intake document flags that has nothing to do with law and everything to do with naming. The cohort numbering — whether the first cohort is Cohort 1 with series 1, 2, and 3, or something else — is recorded as needing confirmation. Do not assign C2 numbers until Cohort 1 is defined. That sounds trivial next to the legal questions, and it is not: a series number is the identity a project carries into every later document, every bank account, and every offering. Getting it wrong at intake means correcting it everywhere later, or living with two names for one company. The gate catches that too, if the people running it are paying attention.

When the first proposal does move through the flow, the test will be simple and public. Intake will file five facts. The Foundation will minute a mission decision with reasons. A master will be assigned and minuted. A series will be designated, a charter signed, a Council seated, a manager named, books segregated, IP assigned. The Council will review and decide by two-of-three signed votes. Execution will follow only the decided step. And a dated receipt will close the loop. Until that receipt exists, the mission gate is a design — a careful, deliberate, well-drawn design, and still only a design. The door is built. Nobody has walked through it yet.

What this is / What it isn’t

What this is

  • An intended seven-stage process — intake, mission-fit review, legal assignment, series designation, Steward Council review, human decision, execution and record — with the Foundation board holding the mission screen and the power to withdraw approval.
  • A story about the brake inside a fast machine: agents draft and prepare at every stage, and humans decide at every stage that matters, with a decline meaning no series number, without exception.
  • A public record of what is blank: no series designated under the process, no Council seated, no Delaware master named, no charter signed, and no proposal yet through the flow.

What it isn’t

  • Not a live pipeline: no proposal has moved through the seven stages, no money has moved under the studio model, and no project has passed any stage gate under it.
  • Not an offering: nothing here is a price, a term, or a promise of return, and no SAFE, equity, or grant has been issued or made under the process.
  • Not a filing: a register row is a tracking row, not a filing, and a series exists only when its master LLC's operating agreement designates it, its books are segregated, and any required designation is filed or recorded.

Where to see it

The operations pages carry the full detail behind this feature: Proposal Flow, Human Gate, and Studio Model, with the governance detail on the Steward Council and the Operations hub. The trueforge-site landing page is public on GitHub: trueforge-site — a landing stub, not a working harness.

Nothing on this page is legal, tax, or financial advice. The intake checklist, the registers, and the seven-stage flow are templates and intended designs. The Foundation is a filed California nonprofit public benefit corporation, and no IRS 501(c)(3) determination letter is in the files.

← Prev: The Agentic Venture Studio · Next: The Steward Council →