Gatsby: We embed builders inside industrial companies and ship the systems that move real operating numbers.
A fixed-price diagnostic maps your operation and ranks the build by expected movement in the number you care about. Then a fixed team, in a fixed window, ships it into production. In your stack. Owned by your people.
01 · The problem
You've already tried this.
A dashboard nobody opens. A pilot that never left the sandbox. A consultant's roadmap in a drawer. The tools were fine. The gap is engineering: someone has to sit with the schedulers and the plant managers, learn the exceptions nobody wrote down, and wire the system into the screens your team already uses.
The distance between a working demo and a changed P&L is an engineering problem, and it cannot be closed from outside your walls. Gatsby puts the builders inside them: two or three people, in your stack, under your security review, building against one operating number we agree on before we start. When they leave, your team runs the system.
See how the engagement runs02 · How we work
Fixed teams, fixed windows, four phases.
Every engagement starts with the diagnostic. What follows depends on what it finds, and each phase is priced before it begins.
01Diagnostic
From $15K · 2 weeks · fixed
Two weeks inside the operation. We interview the people who run it: schedulers, dispatchers, underwriters, plant managers. We map the workflows, cost the pain in hours and dollars, and rank the candidate systems by expected movement in your operating numbers.
You leave with a build plan you could hand to any engineering team. About a third of the time we conclude you don't need us, and we say so.
02Embed
$150–250K · 12 weeks · first system live by week 8, median
Two or three builders on your floor. They ship the highest-ranked system from the diagnostic into production: a forecast on the dispatch screen, risk scores in the underwriting queue, anomaly flags on the line.
One number per engagement. If the work stops moving it, we stop the work.
03Build
$250–400K · 16–24 weeks
Deeper infrastructure: predictive models trained on years of ERP history, document agents that answer from your contracts with an audit trail, integrations across plant systems that have never talked to each other.
Same cadence as Embed, larger scope, still in your repos and under your names.
04Sustain
From $8K/mo · optional
Handover is a phase, not a promise. Before we leave: runbooks, on-call guides, your team trained, documentation that answers the questions we used to.
Most clients run the systems themselves afterwards. About a third keep a retainer for model retraining, new signals, and the next system on the list.
03 · Pricing
How we charge.
The diagnostic is fixed. You know the price and the deliverable before we start. Embed and Build run as fixed teams in fixed windows: scope bends, timelines don't.
The unit of work is a system running in your stack. If it isn't in production, it isn't done, and you never pay for hours nobody can verify. No decks at any stage. The artifact is running software.
04 · The gap
A Tuesday now. A Tuesday after.
Before
- ×Planners override the forecast because nobody trusts it
- ×Nine years of ERP history sit in tables nobody queries
- דWhich contracts do we breach?” takes three days and four people
- ×The exceptions live in one scheduler's head
- ×Every improvement project ends as a slide deck
After
- The forecast sits on the dispatch screen, and override rates are tracked
- Models score risk and flag anomalies from the history you already own
- Contract exposure is one question, answered with citations
- The workarounds are documented and wired into the system
- Your team ships the next system without us
05 · Who it's for
Built for operators with real assets and old data.
Manufacturers, distributors, logistics operators, insurers of physical things. Businesses where the operating numbers live on a floor, in a fleet, or in a book of risk.
- Revenue
- $20M – $500M annual
- Shape
- Plants, depots, or fleets. An ERP, and spreadsheets holding it together.
- Data
- Years of operational history, mostly unqueried.
- Posture
- You want production systems, not another pilot.
If the data isn't there yet, the diagnostic will say so, and we'll stop there.
06 · The work
Systems we ship.
Every operation is different, so every build is scoped from the diagnostic. Here is what one looked like in production, and how an agent engagement runs. Client names are under NDA; references come with the diagnostic.
Regional food distributor
From gut-feel dispatch to a forecast planners trust
Nine years of ERP history held the answer. We built demand forecasts, risk scores, and anomaly flags on the data they already had, and wired them into the dispatch screens the planners already used.
Worked scenario · The Friday line-down
Ask the operation anything, with citations
Language models put your documents on call: contracts, claims, tickets, maintenance histories. Retrieval keeps every answer grounded in the source, and a person signs off where it matters.
You
If packaging line 2 is down until Friday, which customer contracts do we breach?
thinking · Line 2 feeds three SKUs. Check committed service levels per contract, then orders already scheduled this week.
affected_orders(line: "pkg-2", until: "fri")→ 3 SKUs · 11 open orders · 4 due this week.
service_levels(skus: ["405", "417", "442"])→ Two contracts at 98% OTIF · one 48-hour cure window.
breach_exposure(orders: […])→ 2 orders breach Thursday · penalty cap €40k · notice due 24h ahead.
Gatsby agent
Two orders breach Thursday — both under the Meridian contract. Notice is due by
Every answer cites the clause and the ERP rows it came from.
07 · Common questions
What operators ask before they call.
How long until something is live?
The diagnostic takes two weeks. In an Embed, the median time from kickoff to first system in production is eight weeks. The full window is twelve.
What does it cost?
The diagnostic starts at $15K, fixed before we begin. Embed engagements run $150–250K over twelve weeks; Build engagements run $250–400K over sixteen to twenty-four. Sustain, if you want it, starts at $8K per month. Each phase is priced and agreed before it starts.
Do we need our own technical team?
No. Our people build alongside whoever you have, and training your team is part of the engagement. Before we leave, your people run the system and the documentation answers the questions we used to.
What do you actually build?
Demand forecasts, risk scores, and anomaly flags built on the ERP history you already own, plus document agents that answer from your contracts, claims, and tickets with an audit trail. All of it wired into the screens your team already uses.
What if a model can't beat the incumbent spreadsheet?
We say so and stop. It happens. The diagnostic exists so most of those cases never reach a build.
Who owns the result?
You do. We work in your codebase, on your infrastructure, under your names. When we leave, everything stays.
08 · The firm
Integrated engineering.
- 01We work in your codebase, not ours. When we leave, everything stays.
- 02One number per engagement. If the work stops moving it, we stop the work.
- 03No decks. The artifact is running software.
- 04Fixed teams, fixed windows. Scope bends; timelines don't.
- 05We'll say so when you don't need us. It happens about a third of the time.
09 · Start
Tell us what's slow.
One email to start. Bring the process that annoys you most, and we'll tell you whether it's a data problem, a software problem, or not worth solving.