STUD

Every play your startup needs, ready to run by expert agents.

You call the plays. A system that knows your business runs them, and learns from every result.

Intelligence became a commodity. Running a venture got barely lighter.

The hours go to the work around the work: explaining the business again in every new session, carrying each answer to the place it lives, keeping yesterday's numbers from going stale. You were promised leverage and got a second job, operating the intelligence.

It knows your business.

State a fact once and it flows into everything that runs. Your venture, structured and current.

Drafted by a play: a business model canvas, built from stated facts.

Key Partners

We expect to partner with model providers and agent frameworks to source supply, and with communities of practice to grow playbook coverage.

Test: At least 2 agent frameworks agree to a referral test.

Key Activities

We must grow the play catalog, keep workspace facts flowing into every run, and settle results reliably at low cost.

Test: The catalog grows by at least 20 plays per month, each settling against its own contract.

Key Resources

We need the catalog of ready-made plays, the workspace fact store that pre-fills every intake, and the acceptance engine that checks each result before it lands.

Test: A workspace fact set pre-fills at least half of a new play's required intake fields.

Value Propositions

We believe founders will adopt a living system that runs their venture's work from ready-made plays, because writing prompts and reworking chat output burns the hours AI was meant to save.

Test: At least 5 of the first 12 founders interviewed choose picking a play over writing a prompt for the same piece of work.

Customer Relationships

We believe self serve onboarding plus a workspace that compounds will retain them: the more a venture runs, the more its workspace knows, and the better the next run gets.

Test: At least 40 percent of active buyers run a second play within 14 days of their first.

Channels

We expect to reach them through search, founder and developer communities, content, and a self serve web signup.

Test: A community post or search landing page converts at least 2 percent of readers to signup.

Customer Segments

We expect the first customers to be solo and early-stage founders and small product teams who need marketing, finance, and product work done while staying small.

Test: At least 20 founders matching the archetype activate a workspace in the launch quarter.

Cost Structure

We assume model inference and engineering are the main costs, falling per run as plays are reused across many ventures.

Test: Per run model cost stays under 1 dollar at launch volume.

Revenue Streams

We expect buyers will subscribe to monthly credit plans and top up with one-time packs, spending credits on each play they run.

Test: At least 20 percent of triallers convert to a paid plan within 14 days.

It does the work.

Pick a play from the catalog and AI agents run it with your facts already in hand. Ready-made work, checked before it lands.

Computed by a play: unit economics, verdict included.
5Contribution usd
500Contribution cents
trueMargin positive

Breakdown

Take usd
7
Rail cost usd
1.025
Judge cost usd
0.5
Trust safety usd
0.375
Serve cost usd
0.1
Per payee split usd
0
Inputs
Gmv
25
Take pct
0.28
Judge cost
0.5
Ts rate
0.015
Ts fixed
0
Rail pct
0.029
Rail flat
0.3
Serve cost
0.1
Payee count
1
Per payee fee
0
Floor
1.5

Arithmetic recomputed against tasks.py's _unit_economics_reference oracle (the D-009 v2 section 2 contribution identity), 2026-08-17: take is max(0.28 x $25, $1.50 floor) = $7.00 (the floor does not bind); payment rail is 0.029 x $25 + $0.30 = $1.025; trust and safety is 0.015 x $25 + $0 = $0.375; verification (judge) is $0.50; serving is $0.10; single payee, so the per-payee split term is $0. Contribution = 7.00 - 1.025 - 0.50 - 0.375 - 0.10 = $5.00 per run, a 20% contribution margin on the $25 transaction, so each run clears its variable costs before fixed spend. The scenario is ILLUSTRATIVE: STUD is pre-seed with subscription pricing live today ($20 / $60 / $180 monthly plans), and a marketplace take rate on third-party supply is a later phase, so this models that later phase on a $25 third-party play run (this play's own settlement price) at the D-009 anchor 28% take. The takeaway that sets the pricing decision: the $0.30 flat rail fee plus the $0.50 flat judge cost together are $0.80 of value-independent cost, which is why thin runs near the $0.40 to $1.60 play-cost range cannot settle one-by-one on card rails and need batching or the per-task floor.

It learns from every result.

Results land structured, stay current, and feed the next play. The system compounds.

Composed by a play: the same canvas, enhanced from three earlier results.

Key Partners

Supply-side partners that keep producer capacity and inference costs healthy.

  • Model inference providers
  • Agent framework partners
  • Cloud infrastructure vendors

From: Operating Model Canvas: suppliers

Key Activities

The living loop is the business: facts pre-fill the intake, the play runs, the checked result lands back in the workspace.

  • Pre-fill the play intake from workspace facts
  • Run the play with an AI agent
  • Check the deliverable against the play's contract
  • Weekly operating cadence reviews

From: Operating Model Canvas: value delivery chain

Key Resources

The workspace record, the play catalog, and the acceptance ledger are the assets a venture relies on.

  • Workspace facts record
  • Catalog of ready-made plays
  • Acceptance event ledger
  • Vercel and Neon cloud infrastructure

From: Operating Model Canvas: locations, organization, information

Value Propositions

A living system for the venture: ready-made plays run on workspace facts, and every accepted result makes the system smarter.

Products & Services

  • A catalog of ready-made plays run by AI agents
  • A workspace fact store that pre-fills every intake
  • Structured deliverables checked before they land

Pain Relievers

  • Plays replace prompt writing with picking
  • Facts stated once flow into every run
  • Results land structured and stay current

Gain Creators

  • Each accepted result feeds the next play
  • The workspace compounds the venture's knowledge
  • Checked delivery makes results safe to use immediately

From: Value Proposition Canvas

Customer Relationships

Self serve onboarding with a workspace that compounds: the more a venture runs, the more the system knows.

Channels

Self serve web signup reached through founder and developer communities and content.

Primary: Self-serve-web

Customer Segments

Solo and early-stage founders and small teams, prioritized by their need to get real work done while staying small.

Jobs

  • Get real work done while staying small
  • Keep the venture's knowledge current

Pains

  • Every AI session starts from a blank prompt
  • The same company facts get re-explained everywhere

Gains

  • Work that starts from what the venture knows
  • A workspace that gets smarter every run

From: Customer Profile

Cost Structure

Costs concentrate in inference and the platform team, falling per run as plays are reused.

  • Model inference spend per settled run (driver: Model inference providers)
  • Platform engineering payroll (driver: Founder-led product and engineering)
  • Cloud hosting and data (driver: Vercel and Neon cloud infrastructure)

From: Operating Model Canvas

Revenue Streams

Buyers subscribe to monthly credit plans and spend credits on each play they run.

Primary: Pay-per-use

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

Opens soon. Get early access the day it goes live.

You wear every hat. This week needs marketing, finance, and legal attention, whether or not you have done those jobs before. Nobody knows every job. The catalog exists so that you never have to.

One quiet property holds it all up: every deliverable passes its play's acceptance check, a contract fixed before the work begins, before it is allowed to land.

So make it so.

Browse plays