The Protocol · Ninety Days

Register.
Run. Read.
Publish.

A trial is a protocol with the order fixed in advance, and the order is what makes the result mean anything. Every week produces a real artifact. On day one, founders already know the thresholds that will decide day ninety.

00The protocol

Four steps, in this order.

01
Register

Before the clock starts, the criteria go on paper: buyer list, wedge, and the thresholds that produce each verdict. Criteria set after the data arrives are not criteria.

02
Run

Ninety days. Thirty to build, sixty to sell. No curriculum, no demo day, no audience but the buyers. The studio carries the build and the entity so the founder can sell.

03
Read

On day ninety the registered thresholds are applied to real billing data and a verdict is issued: GO, GROW, PARK or KILL. The verdict is computed, not negotiated.

04
Publish

Every verdict goes on the record, including the ones that embarrass us, and every verdict is tracked to month eighteen in the calibration ledger.

One vocabulary note. The day-ninety output is a verdict. The month-nine durability test, which asks whether the customers stayed, is a verdict. They answer different questions and we do not let one stand in for the other.

01Weekly ledger

One artifact a week.

  1. 01
    Week 1
    Lock the market and the contract

    We confirm the target account list and sign a short operating agreement covering what ships, what gets reviewed, and where the decision points sit. Access founders bring their own market and we pressure-test it. Technical and Match founders take a scoped wedge from the studio.

  2. 02
    Week 2–3
    Ship a first version to a real account

    Working software in front of one paying-conversation account by the end of week three. Deployed to production, with a login, running against a real workflow. No prototypes.

  3. 03
    Week 4
    First letter of intent or paid pilot

    Written intent or a small check from the named buyer. If nothing lands, we recut the market rather than push a weak product into a bigger pipeline.

  4. 04
    Week 5–8
    Turn on the sales team

    The studio's sales bench works the target account list: outbound, discovery, pilot conversion. The founder stays in every buyer conversation. Target for the phase is the first paying customers on the path to $100K annualized by Day 90.

  5. 05
    Week 9–10
    Draft the pre-seed package

    Deck, metrics, cap table, security posture, references, model. Everything a pre-seed investor asks for on the first call, assembled before the first call happens.

  6. 06
    Week 11–12
    Open introductions

    Introductions to pre-seed funds and angels who write $250K–$1M on a $4–6M SAFE cap. Few good conversations beat a wide net. The founder takes the meetings.

  7. 07
    Week 13
    Day-90 verdict

    Every company resolves to one of four verdicts: GO, GROW, PARK, KILL. Computed against the thresholds registered at day zero, written down, and communicated to the cap table the same week.

02Day-90 verdicts

Four honest endings.

Two of the four are wins: the company raises, or the company generates real revenue. One is patience. One is an honest ending. The industry counts unicorns. We pay closer attention to the number of companies still operating a year later, run by the founders who built them.

GO
Pre-seed ready

The evidence supports a pre-seed on a $4–6M post-money SAFE, typically $500K–$1.5M. We open the introductions. The founder runs the process and closes. Seed comes later, off a $300K–$500K ARR base.

GROW
Commercial, seed comes later

Real revenue and slope, but not yet the base a seed fund will lead. Compound another quarter to $300K–$500K ARR and open seed introductions, or run for cash flow. Both are wins.

PARK
Signal early

Something worth watching, but the market or the moment isn't ready. The studio moves to a lighter cadence and sets a specific date to look at it again.

KILL
Negative verdict

Rare. Reserved for the case with no traction and no path. The founder keeps everything: code, customers, domain, name. Nothing gets clawed back.

03Per-company mechanics

How each company is run.

Cohort size
3–5 companies at a time. Small enough that every founder gets partner time every week. Large enough that the sales work compounds across the group.
Operating cadence
One ship review a week, one pipeline review a week, one monthly metrics readout to the cap table. No office hours in perpetuity.
Founder development
This is an apprenticeship with a company attached. Weekly ship and pipeline reviews sit next to the work rather than above it, investor prep starts Week 9, and the founder is in every buyer conversation. By Day 90 the founder has scoped a wedge, shipped it to production, and closed price without a partner in the room.
Build stack
AI-first tooling on the studio's infrastructure and credits. A technical partner if you're solo, or a matched founding CTO if you're an operator. Every account and repository sits in the company's name from day one.
Sales
The sales bench runs outbound, discovery, and pilot conversion against the target account list. The studio owns the CRM. You own the buyer relationship and the close. The bench is paid in studio equity, never billed to the company.
Build stack cost
Everything the studio supplies is paid in studio equity. Direct founder cash outlay across the ninety days is roughly $5,000. The fully-loaded studio cost of the build, about $150,000 at market, becomes the SAFE principal. Credits stack up to $500K per company; a company usually draws $50–75K.
Standard docs
Delaware C-corp, one class of common plus our preferred stake. Studio agreement, IP assignment, founder stock docs. We do not sign NDAs. Everything templated, counsel-reviewed, signed before day one.
How a cohort is measured
One number. Verdict rate, defined as GO plus GROW. A cohort where one company raises and two others reach real revenue is a good cohort.
Where we work
Distributed. Founders work where they already live, anywhere the hours overlap the weekly cadence. We sync on a real cadence instead of a mandatory office.
Cohort 01 · Applications open

Ninety days.
One verdict.