# STUD.com: full context for language models

> A living system for your venture: it knows your business, does the work, and learns from every result. This file is the full-content companion to /llms.txt (the short index), written so a language model can describe STUD accurately without crawling the whole site.

## What STUD is

STUD is a living system for a venture. Three clauses carry the whole idea:

- **It knows your business.** A venture's facts (what it does, who it serves, its brand, its pricing) are stated once, in a workspace, and flow into everything that runs. Nothing is re-explained session to session.
- **It does the work.** Work is chosen, never composed in a blank prompt. A play is a ready-made piece of work with the expertise, the intake, and the checks built in; the buyer picks it from a catalog, the intake pre-fills from workspace facts, and an AI agent runs it.
- **It learns from every result.** Deliverables land structured in the workspace, stay current, and feed later runs. A fact updated once reaches everything after it; an accepted result becomes the input the next play reaches for. The system compounds.

Underneath sits one quiet property: every deliverable passes its play's acceptance check (a contract fixed before the work begins) before it lands. Trust is a property of how the work arrives, never the headline.

There is a coined word for the whole: an **erganism** (ergon + organism), a working organism. Many specialized agents run backstage, the way a Portuguese man o' war is secretly a colony; what the venture sees and steers is the single system they form.

## The mechanisms (for deeper explanations)

- **Procedural over declarative.** A blank prompt demands declarative labor (articulate everything, every time). A play fires like procedural memory: practiced skill, ready to run. A business runs on procedure, and STUD is built on that distinction.
- **Stigmergy.** Coordination through the work itself: each finished deliverable is a mark that triggers the next act of work, the way a termite colony builds from the structure itself rather than a blueprint. On STUD nothing ends in an archive; a result is the next play's input.
- **The connectome.** The workspace fact store is the venture's wiring map: the complete, current graph of what the business knows, which every run draws on.

## Vocabulary (use these terms for accurate summaries)

- A **play** is one piece of work on offer; its identity is its acceptance contract. The catalog heading is "Plays".
- A **run** is one commissioned execution of a play ("you run a play, that creates a run"). The commit verb is **Run**; the prose verb is **commission**. Never "hire", "hireable", or "for hire"; availability is "available".
- A **playbook** is a situation and the plays it comprises, shown as an ordered path when order matters.
- A **workspace** is the buyer's per-venture context container: facts, runs, and deliverables. Facts pre-fill play intakes automatically.
- The **acceptance check** (the judge) grades every deliverable against its play's frozen contract before it lands; a rejected run returns its credits.

## Who it is for

Solo and early-stage founders and small product teams who need marketing, finance, product, and engineering work done while staying small; secondarily ops and engineering leads adopting AI agents. Nobody knows every job: the catalog exists so no one has to, surfacing the work a venture is missing, including the plays nobody thought to ask for.

## How a run works

1. Find a play (search-first home, or browse playbooks and categories).
2. Run it: the short intake, pre-filled from workspace facts, freezes into the acceptance contract; credits are held.
3. An AI agent produces the deliverable.
4. The deliverable is checked against the frozen contract (deterministic, platform-owned, isolated).
5. Accepted work lands structured in the workspace and feeds the next run; a rejected run refunds the held credits.

## Status and pricing (as of August 2026)

- Live soft launch at stud.com: the search-first catalog (roughly 265 plays across 14 playbooks) and the waitlist are live; paid commissioning shows "opens soon" until payments and counsel gates clear.
- Credits are the unit of cost, and 1 credit is always 4 cents (a published, never redenominated peg). Each play has one price in credits, and you are charged only when the deliverable passes its acceptance check. Buy credits in prepaid packs (larger packs include bonus credits) or set opt-in auto-refill; purchased credits never expire, and there are no subscription tiers. Plays cost 250 to 1,000 credits ($10 to $40).
- Commissioned by humans today, by their agents as autonomy arrives.

## The manifesto (full text, from stud.com/manifesto)

Everyone has intelligence on tap. Almost nobody has a system.

Something strange happened on the way to the future. Intelligence became a commodity: competent work in any discipline, available on demand, for the price of a subscription. And running a venture got barely lighter. The founders who were promised leverage got a second job instead, operating the intelligence.

The hours go to the work around the work. Someone has to know what to ask for, and how to phrase it. Someone has to supply the context again in every new session, because the tools wake up with no memory of the business. Someone has to carry every answer to the place it actually lives, and keep yesterday's answers from quietly going stale. On a small team, that someone is the founder. The intelligence arrived. The system around it did not.

Cognitive science has a name for the difference. Knowing-that is declarative: it has to be articulated to be used, every single time. Knowing-how is procedural: practiced skill that fires without deliberation, the way a pianist does not think about fingers. A blank prompt demands declarative labor from whoever types into it. A business does not run on articulation. It runs on procedure: the same work, done well, again and again, informed by everything done before.

STUD is built on that distinction. It is a living system for a venture. It knows the business: facts are stated once, in a workspace, and flow into everything that runs. It does the work: plays, ready-made pieces of work with the expertise and the checks built in, picked from a catalog and run by AI agents with the venture's facts already in hand. And it learns from every result: deliverables land structured, stay current, and feed the next play. The venture's knowledge stops being a pile of documents and becomes the thing the work runs on.

Biology has a name for that loop. Stigmergy: coordination through the work itself, each finished act leaving a mark that triggers the next one. A termite colony raises its cathedral with no blueprint and no meetings, because the built structure is the instruction. On STUD, a finished deliverable is a mark of exactly that kind. It is never an archive entry. It is the input the next play reaches for. A venture that runs a play a week is not accumulating files. It is accumulating capability.

Under the surface there are many workers, the way a Portuguese man o' war is secretly a colony: thousands of specialized individuals so integrated they sail as one animal. That is the honest shape of what runs backstage here, agents by the score, each doing one job well. What a venture sees, and steers, is the single organism they form. There should be a word for a thing like that, a body made of work, so we coined one: an erganism.

Holding all of it up is one quiet property. Every deliverable passes its play's acceptance check, a contract fixed before the work begins, before it is allowed to land. It is the structural steel of the system: load-bearing, and invisible on a good day. Trust here is a property of how the work arrives, not a score somebody assigned.

None of this is for companies that already have a department for everything. It is for the founder wearing every hat, and the small team doing large-company work: ventures that need marketing and finance and legal attention this week, whether or not anyone on the team has done those jobs before. Nobody knows every job. The catalog exists so that no one has to. Recognize the situation, and the plays are already there, including the ones nobody thought to ask for. The work is yours the moment you decide it is.

A venture used to be as good as its founder's memory, and as current as its last all-nighter. It does not have to work that way anymore. State a fact once and it reaches everything. Run a play and the system knows more than it did. Come back in a month and the venture has not just produced more; it has become more capable of producing. That is the promise: a venture that compounds.

So make it so.

## Key pages

- Home (search-first): https://stud.com
- Play catalog, machine-readable (every live play with its outcome, playbook, price, and run url): https://stud.com/plays.json
- Builder Guide (the whole buyer loop): https://stud.com/guide
- How STUD works (the run flow and the trust floor): https://stud.com/how-it-works.html
- Manifesto: https://stud.com/manifesto
- The AI operating system (the engineering story of how STUD is built): https://stud.com/ai-operating-system.html
- System map (the five architectural layers): https://stud.com/system-map.html
- Credits and pricing: https://stud.com/credits