Closed-loop · Detroit

A closed-loop agentic platform. We start companies on it.

We could have sold the loop. We form companies on it instead — stay until they run, then the seat fills. Investors buy that company.

Why a studio

We didn't unleash the platform. We build companies on it.

01

The loop

Agents, orchestration, memory, a harness that decides when to act. Closed.

THE RUNTIME →

02

The company

We form it on that loop. First code, first sale, in an industry that passes.

THE FILTER →

03

The seat

When it runs, an operator takes it. The loop stays. Investors buy the company.

THE FOUNDER SEAT →

Investors come in at the company, from the beginning. How investing works →

The industries

Rules in the path. Or a number large enough to matter.

A regulator, an auditor, a court — or a pool of spend that can carry a company. That's how we choose where to build.

Regulated

A regulator, an auditor, or a court is downstream of the work. Rules create the document trail — and the barrier that keeps a general-purpose tool out.

Document-heavy

The work arrives as paper, faxes, PDFs, spreadsheets, contracts, claims. Unstructured input is the reason humans were doing it.

Fragmented

Thousands of small operators, no dominant vendor, no shared standard. Nobody has the scale to build their own way out.

Cost scales with volume

Every extra transaction costs another unit of human attention. Growth is punished. This is the test that matters most.

Big enough to matter

The pool of operating spend has to justify a company. Market size is the last test, not the first — a large market with none of the above is not our problem to solve.

Companies we've started
on the closed loop.

Healthcare, licensed retail, music rights, property, documents. Different industries. The scorecard is the better way to read them.

Currently

Health information exchange · Healthcare data

Clinical data plumbing — HL7 to FHIR to OMOP, on-premise, local models, full audit trail.

$740B

U.S. healthcare admin spend / year

●●●●●

CNBS

Regulated retail operations

An operating system for licensed retail in one of the most fragmented compliance regimes in the country.

$30B

U.S. legal cannabis retail / year

●●●●

Kora

Music rights · Africa

Royalty recovery and rights administration for artists whose work earns most of its money outside the continent it came from.

$13B

Global collections. Africa’s share is 0.7% — the money is abroad

●●●●●

ZoomProp

Property operations

Deal and transaction intelligence for an industry where the paperwork is the product.

$1.6T

U.S. existing-home sales / year. The paperwork is the product

●●●●●

ClearInfo

Document intelligence · Diligence

Cited answers over document sets nobody has time to read end to end.

$400B

U.S. legal services spend / year — the work is in the documents

●●●●●

Colony

Sales automation · Workspace

Agent-assisted outreach, follow-up, and workspace automation on the same operating model as the rest of the studio.

72%

of a sales week is not selling — CRM, notes, follow-up

●●●●●

Multiplying

Personal AI · Voice

Talk for 30 minutes. We clone your voice and how you think — then you gate and price the knowledge.

$250B

Creator economy. Expertise still sells by the hour

●●●●●

Buy the company. Not a fund.

Money goes into the company itself, from the beginning. The terms are published.

How investing works →

Filled circles are tests passed. The numbers are the market, not the company — Whitney, CISAC, NAR, and the thesis page.

The loop

The platform is the business. The studio is how we deploy it.

Every company we start inherits the same two things. The first is the operating layer — the agents, the orchestration, the memory, the evaluation harness that decides when a machine is allowed to act and when it has to stop and ask.

The second is harder to copy and took longer to build: the posture around it. HIPAA. Business associate agreements. Zero-trust architecture. Models that run on the customer's own hardware because the data is not allowed to leave. An audit trail that survives contact with someone whose job is to find the gap.

We built that for hospital data — the hardest regulated environment in American software. Everything since sits on that same bar. A collecting society asks for less than a hospital infosec review. So does a title company. If it clears the hospital, it clears the next one.

Agents are easy to buy. Passing a hospital review on day one is not.

What every company starts with →

What happens to the company

The loop stays. The seat fills.

First customers, first workflows, the loop underneath. When the company can stand, an operator who knows the buyer takes the seat — their board, their name on the door.

What we keep is the closed loop. Each company teaches it something, so the next one does not start from zero. That is why we are a studio, not a vendor.

What we don't do

No consulting. No studio fund. No invented numbers.

We only build for companies we started.

No consulting. No agencies of record. No advisory retainers. We build for the companies we started, and the arrangement is published on the deal page.

We don't raise at studio level.

Midwest is carried by its founders. Money is raised into the companies, by the companies, from investors who are buying that company — not a piece of Midwest.

We don't publish numbers we can't stand behind.

There are figures on this site marked as targets, and figures marked as measured, and we do not mix them. Where we don't know yet, it says so.