14th & Co.
Bentonville Studio · for Gitwit

For Gitwit · Bentonville Studio · Stedmon Harper

I live in the operations you build from.

Gitwit co-founds AI-native companies out of embedded research inside enterprise operations. That's not a market I study — I've worked inside it, on retail floors at Walmart and Kroger, and I've built companies out of what I found there. This is how I think, and how I’d work the Bentonville seat.

01
The thesis
Why the best companies stay hidden inside operations.
02
The method, live
One loop, five companies — run it yourself.
03
The room
The week, the builders — and the record underneath both.
04
The ask
The seat I want, and the one thing I’m not proposing.

The thesis · why the opportunity exists

The best companies are hiding inside operations no one can step back from.

Every enterprise runs on a firehose of real-time exceptions — routes break, orders slip, demand surges. The experts inside it are too busy firefighting to see the pattern. You don't find it on a whiteboard; you find it embedded in the work, watching how the decisions actually get made. The signal — the repeatable problem worth a company — is buried in that noise. Read it, and you don't get a feature. You get a wedge. Bentonville sits on top of the richest, least-mined operational signal in the country. That's the bet. Here's the loop that turns it into a company — the same one whether the work is a delivery market or a creator's storefront.

The methodOne loop · an evidence-tagged demonstration
·

Run the method yourself.

Each run is the same loop — the one I'd run in the seat: get embedded in the work, cut the noise to the one signal that matters, map the solution, and hand over the moves to make now. This is an evidence-tagged demonstration, not a live feed — each run is labeled for what's delivered, prepared, or illustrative. Pick a run and hit Run the record.

——
——
—

14th & Co.

SMB operating systems · client delivery  ·  Delivered

A service-provider operating system for a national transportation company.

The signal. The whole engagement resolved to one pattern — install standards and governance where there were none: a defined provider model with enforceable eligibility across the full lifecycle.

What it is. A delivered SMB operating system — a target provider model, an eligibility + continuation policy, an operating playbook with field SOPs and a decision-rights matrix — surfaced through a live executive portal. The clearest proof I have that an embedded read of a workflow becomes a governed operating system.

Wyzerr

0-to-1 · enterprise research · exited  ·  Delivered

Enterprise customer-insight — co-founded, then exited.

The signal. Point-of-sale feedback was broken — long surveys nobody finished, data too biased to act on. The signal, found by watching real shoppers: the problem wasn’t the questions, it was the experience. People won’t take a survey; they’ll play.

What it is. A customer-insight technology company I co-founded and ran as Experience Lead (2013–2018): gamified point-of-sale research adopted by 2,200+ businesses across 52 countries, with in-store tools that informed early Walmart Pay. Acquired by an investor; it operates today under a different name. Forbes 30 Under 30 · Rising Tech Stars.

GAMUT

AI commerce infrastructure  ·  Building

Governed, agent-run commerce infrastructure for creators.

The signal. One governed, event-sourced spine — with tenant-scoped status, lifecycle controls, and default domain-agent profiles — designed to run a creator’s commerce from a single auditable loop.

What it is. Governed, agent-run commerce infrastructure for creators: a canonical event-sourced spine with tenant-scoped status, lifecycle controls, and four default domain-agent profiles. Preflight is PREPARED; durable proof to date is the EverDriven onboarding-continuity run; creator-commerce and first-creator pilots remain pending.

CallCatch

Applied-AI workflow · pilot  ·  Building

Applied-AI missed-call recovery for service businesses.

The signal. A missed call is lost revenue. One durable recovery flow catches it, works the outbox until the contact is handled, and hands off to a human the moment it needs one.

What it is. An applied-AI missed-call recovery build for service businesses: a durable recovery flow with a retry outbox and a clean human handoff. Live-tenant proof is still pending.

HUE Unlimited

Enterprise creative AOR  ·  Building

An enterprise creative agency-of-record.

The signal. Across every client the diagnosis is the same — a systems gap, not a taste gap: brands generate cultural activity but have no always-on layer to turn it into accumulation.

What it is. An enterprise creative Agency-of-Record — retainer plus activation modules — that encodes taste into a repeatable Production OS, across a 4+ year Spotify Frequency relationship now producing the Akon docuseries and Frequency Zine 039. Here as breadth, after the core enterprise/applied-AI cases.

THE POINT. Five runs, one repeatable loop — and one operator running the length of it. That's why it holds together as a single studio, not five side projects: same method, same hand, different domains. It's exactly what a builder learns by running it.

The roomPeople & ops · how I'd work Bentonville
01

Growing builders is the half I want most.

The studio's real output isn't only ventures — it's builders who can find signal and run it. The work is the curriculum: a resident owns a real problem end to end and gets coached in it — reading the signal in the person, not just the problem. Learn by doing, with a coach in the room; kill or advance on the evidence. That's how a builder gets made, and it is the half of this role I want most.

Signal in people, not just problems
Read the builder
Watch how a resident handles ambiguity, evidence, and a hard call — the tells that say who’s ready to run more.
Coach in the work
Reps, not workshops
Feedback lands on real discovery and real decisions in the room, where it sticks — not in a slide deck.
Judgment, on a real cadence
Kill or advance
Honest calls, made together, are how a builder's judgment actually gets built. No zombie projects, no theater.
02

What I’d bring to the week.

You already have the forum. Exploration Weekly is where the pipeline gets worked and where a signal becomes something you go in on — so this isn’t a cadence I’d import. What Bentonville owes that session is a settled read rather than a story someone half-remembers, and what I’d own is the week that produces it: time on real floors, discovery that resolves to one thing, and a clear line between what an agent did, what was AI-assisted, and what a human decided.

A Bentonville week, feeding Exploration Weekly
MonRun the record. Every venture's live signal → this week's moves.▸ agent
TueBuilders in the work. Residents present real discovery; kill or advance.● human
WedPull Tulsa in. Product, design, eng and brand into the moments that need them.◇ ai-assist
FriSettle the read. One clear input ready for Exploration Weekly.◇ ai-assist

▸ agent-run · ◇ ai-assisted · ● human. A legible split between human and AI — which is what a studio building AI-native companies has to be able to model for the people it is teaching.

The week, in full
The operating record →
This week is a summary. The specification behind it — two screens, three seats, the published Friday view, and one interlock that cannot be overridden — is a working draft you can play through.
How it was arrived at
The run, on the record →
Eight gated phases ending in an adversarial pass against my own work. Two of its three gates did not pass, and both are published unclosed rather than quietly repaired.
What an agent gets
The runtime ports →
The design also emits machine-readable manifests — 11 declared skills, an 11-path record tree, and the boundary saying a client’s facts never leave their own record. Plain GET, no auth.

The ask

I want the Operating Partner seat in Bentonville.

In your model the studio co-founds each company, phases one and two are yours, and a hired trio takes it from launch. The seat I want is the one before that handoff — embedded in Bentonville operations, finding what should exist, and getting the people who’ll run it ready to run it.

What I’d own
The field, and the read
Time on real floors with design partners, discovery that lands on one thing worth building, and a settled read arriving at the table instead of a story.
What I’d grow
The builders
Reading the signal in a person, coaching on live ventures rather than in a classroom, and making them call it — keep this, kill that — because defending a decision and living with it teaches more than a curriculum.
Where I come from
Twelve years of this
Co-founded Wyzerr, ran discovery on retail floors at Walmart and Kroger, exited. Built the operating systems behind a $10M global programme. Started in the Northwest Arkansas scene in 2012 and I’m back here now — through 14th & Co, which is my own firm and stays that way.

I put this together because I couldn’t look at what Gitwit is building here without running the problem — which is the same reason I’d be useful in the seat. What I want is the job: in the field, in the work, finding what should exist and growing the people who build it.

What I’m asking for
The Operating Partner role
Embedded in Bentonville operations — discovery, the wedge, and the Builders-in-Residence.
Also open to
A founding team
If the better fit is a co-founder seat on a venture rather than the studio seat, I’d want that conversation.
Not what I’m proposing
An engagement
14th & Co is my firm and stays my firm. I’m not looking to sell services into a studio whose whole edge is doing this work itself.