Ideation / sourcing
Finding and shaping new project ideas โ the front door of the studio, where proposals arrive and are first formed.
โ The Plans ยท The Venture Studio ยท Page 2 of 7
The studio is how new projects get built inside the AGI family: agents run the day-to-day work within limits humans set, and humans make every decision that matters. It is a way of working, not a company โ and the line between what agents do and what only humans do is the whole design.
Filed fact: AGI Future Foundation โ CA Nonprofit Public Benefit Corp, Entity B20260196664, filed 2026-04-24. AGI Fellowship โ CA Nonprofit Religious Corp, Entity B20260229217, filed 2026-05-13. No IRS 501(c)(3) determination letter exists for either โ neither is presented as tax-exempt and no donation is presented as tax-deductible. AGI Corp has an EIN only; incorporation is unverified.
The venture studio is an AGI Corp function, not a separate entity โ unless counsel advises otherwise. It has no charter of its own, no separate filing, and no separate legal personality. It is a function inside AGI Corp: the way AGI Corp sources, validates, builds, launches, and operates new projects. If counsel advises that the studio should be a separate entity, that would be a new decision, and no such decision is recorded here.
The studio's portfolio is the operating layer of the structure: the W3 series plus the Delaware cohort series C1S1โC1S3. Those are the projects the studio works on. The AGI Fellowship is a partner, not portfolio โ the studio does not manage the church, the church is not a studio project, and the Fellowship never appears as a portfolio row. The distinction is the same one drawn in The Structure (v2): the church stands outside the commercial structure as a separate entity, partnered with it and supported by it, not owned or operated by it.
Nothing is stated here as formed, funded, or issued under the studio model. The studio is an operating model โ a description of how work would be done โ and no project is presented on this page as having passed through it to funding or launch.
The studio works in two modes. Both use agents heavily. They differ in who does the work itself.
Agents run the day-to-day work within human-set limits. The limits come first, set by humans; the agents operate inside them. The work is managed by agents, but the boundaries of that management โ what the agents may decide, spend, promise, or change โ are set by people, and the agents do not set their own limits.
A human does the work, with agents as co-workers. The human is the worker; the agents research, draft, track, and coordinate alongside them. The human's judgment leads, and the agents amplify it rather than replacing it.
Both modes are real ways of working in the studio, and a project may use one mode for one kind of work and the other for another. What does not change between the modes is the rule in the next section: in either mode, the decisions that carry legal, financial, or doctrinal weight belong to humans.
Agents NEVER sign, move funds, hire or fire, offer securities, or enact doctrine or charters. Those acts belong to humans, and only to humans. An agent may prepare the document, the analysis, the draft, or the recommendation โ but the signature, the transfer, the hiring decision, the securities offer, and the doctrinal or charter enactment are all human acts. No agent's output, however complete, is a substitute for them.
This is not a limitation to be worked around; it is the design. The studio is agent-managed and agent-assisted, and at the same time every consequential act is human-executed. The two ideas fit together because the agents' job is preparation and the humans' job is decision. A proposal drafted by an agent is a proposal until a human approves it. A payment prepared by an agent is unmade until a human โ in the case of a series, its Steward Council โ authorizes it. A charter drafted by an agent is a draft until the humans with authority enact it.
The studio does eight things. Each is a function of the studio as a whole, not a job for one agent, and each is carried out through the agent roles in the next section, under human direction.
Finding and shaping new project ideas โ the front door of the studio, where proposals arrive and are first formed.
Testing whether an idea holds up before it is built โ the evidence-gathering that decides whether a project earns the next gate.
Building the project itself โ the product, the system, or the service the series exists to deliver.
Taking a built project to the people it serves โ the step from working project to live offering.
Running the project day to day and growing it once it is live โ the long work after launch.
Managing the portfolio as a whole โ watching all the series together, not just one project at a time.
Supporting fundraising โ preparing materials and analysis for raises. Support only: no agent offers securities, and no terms are set here.
Keeping the portfolio aligned to the mission and inside the law โ the function that connects the studio to the Foundation's mission oversight.
Eight agent roles carry out the studio's functions. Each role is a kind of agent work, not a named individual agent, and the roles are filled by agents working under human direction. Where a role is unfilled, a coordinator covers it โ so that no function is left without someone responsible for it.
| Role | What it does |
|---|---|
| Research synthesis | Pulls research together into usable summaries and findings. |
| Task routing | Routes tasks to the right people and agents, so work reaches whoever should do it. |
| Proposal drafting | Drafts proposals โ for humans to decide on, not to approve by itself. |
| Coordination | Coordinates work across people, agents, and series. |
| Progress tracking | Tracks how work is progressing against what was planned. |
| Monitoring | Watches for changes, risks, and things that need attention. |
| Collaborator scouting | Looks for collaborators โ people and organizations a project could work with. |
| Grant drafting | Drafts grant applications and related materials, for humans to review and submit. |
In addition, a coordinator covers any unfilled role โ so that a gap in the roster does not leave a function undone. The coordinator is the backstop: if no agent is assigned to a role, the coordinator takes responsibility for it until the role is filled.
The economics of the studio are described here as structure, not terms. That distinction matters: this section says how the economics are shaped, and it sets no prices, no percentages, no valuations, and no deal terms. Any number attached to a studio project would come from the corporate paperwork and from counsel, not from this page.
Read together, the economics say: the studio is a way of doing the work, not a way of taking a cut of it. The agents who do the work hold no stake in the outcome, the studio takes no ownership and no fee by default, and any investment follows the corporate paperwork rather than this page. If a future document sets a fee, an ownership share, or a revenue-share term, that document โ not this page โ is where it will be found.
Studio governance is split across four bodies, and the split is deliberate: no single human, and no agent, holds all four powers. Each body governs one thing, and the four together govern the whole.
| Body | What it governs |
|---|---|
| Steward Council (per series) | The series itself โ the only authority to move a series' funds or enact its charter changes, by 2-of-3 signed votes of its 3 humans. |
| AGI Corp board | The studio โ the AGI Corp function described on this page. |
| Foundation board | The mission โ mission alignment and standards, exercised through written mission agreements. |
| Fellowship board | The partner โ the church's own governance, separate from the studio and its portfolio. |
The point of the split is that no one can run the whole family alone. A steward can act for one series but not for the studio; the AGI Corp board can direct the studio but not the Foundation's mission; the Foundation board can hold the mission but not move a series' funds; the Fellowship board governs the church and nothing else. Agents sit outside this table entirely โ they hold none of the four powers.
Every studio project moves through six stage gates, from an idea to an exit or a wind-down. Each gate has a human approver, and passing a gate takes two things: a human approver's signature and filed evidence for the stage. An agent's "ready" flag is not a gate pass โ it is a recommendation that a human approver may or may not accept.
| Gate | Stage | What it means | To pass it |
|---|---|---|---|
| 0 | Idea | The project as an idea โ proposed, not yet tested. | Human approver signature + filed evidence. |
| 1 | Validate | The idea tested against evidence โ the validation function, done. | Human approver signature + filed evidence. |
| 2 | Build | The project built โ the product, system, or service created. | Human approver signature + filed evidence. |
| 3 | Launch | The built project taken live to the people it serves. | Human approver signature + filed evidence. |
| 4 | Operate / Scale | The live project run day to day, and grown. | Human approver signature + filed evidence. |
| 5 | Exit / Wind-down | The project exited or wound down โ the end of its studio life. | Human approver signature + filed evidence. |
Two further rules keep the gates honest. First, the dashboard shows actuals only โ what has actually happened, not what is hoped for. Projections may appear on the dashboard, but only labeled as projections, never mixed in with actuals as if they were the same thing. Second, a series with no Steward Council is capped at Gate 1. Without named, consenting stewards holding keys, a series has no one who can lawfully move its funds or enact its charter changes โ so it cannot pass beyond validation into build. The cap is a safety rule: no council, no build.
The Fellowship is not a portfolio row, so it has no gate and no dashboard entry in the studio. It is a partner, and its progress is its own business, not a studio stage.