Diagnostic, build, care. In that order, on purpose.
Every engagement runs the same path: a two-week diagnostic that proves the build on your real data, one module built to production at fixed scope, and the monthly care that keeps it running and getting sharper. You own the code from day one.
Step 1 · Loop Diagnostic
Two weeks. Fixed fee. A straight answer.
Before I build anything, I prove the build is worth making — on your data, not a slide. The diagnostic is $6–8k, fixed, and the fee is credited toward the build. If the data can’t support the build, I’ll tell you so. That’s what the two weeks are for.
Four jobs, in order
- Map the data. Where creative, spend, and inventory signal actually lives across your stack. Every tool you run stays — each one becomes an input.
- Prove the pieces connect. With your real data, I show that the connections the build depends on actually hold — that your ad results can be tied back to the creative that drove them, that your sales pace can inform what you reorder.
- Back-test against your own history. What would the system have told you last quarter — checked against what actually happened.
- Lock module-one scope. You exit with a fixed-scope, fixed-price proposal for the first build.
Step 2 · Module build
One module, fixed scope, built to production.
We build the workflow the diagnostic proved out — one module, scoped exactly, priced exactly. Fixed fee against the agreed scope, quoted exactly in the diagnostic proposal; most module builds start around $10k. It ends when the system is running in production, documented, and stabilized — not when a prototype demos well.
What production means here
- Staging before production. Nothing ships untested.
- Every change tested automatically before it goes live, with a full history you can roll back to.
- Heartbeat monitoring on every scheduled run; every failure captured, classified, and routed.
- Documentation and runbooks, in your repo, written as we build.
- A client portal showing what ran, what’s queued, and what’s waiting on your approval.
Step 3 · Loop Care
Keeps running. Keeps learning.
Production software isn’t fire-and-forget. It runs on infrastructure that needs tending, and it gets sharper only if someone feeds the corrections back in. Loop Care is the monthly retainer that does both.
What’s included
- Hosting and infrastructure, managed and monitored.
- Maintenance and tuning as your business and the models change.
- A monthly working session: what ran, what it got wrong, what we build next.
- A memory layer in your own database that learns from your team’s corrections, so month six is sharper than month one.
Ownership
The work is yours at every stage.
From the first commit, the system lives in a git repository you own: the code, the database, the prompts, the docs. Most teams have me run it on managed infrastructure — they get the results, and keeping it running, monitored, and current is on me. If your team operates its own infrastructure, I deploy into it instead.
Either way, if we ever stop working together, you keep a running system. Full handoff: source, database export, runbooks, access transfer. What you’d lose is the hands that keep it sharp and the gains that come from staying with it — never the system itself.
Pricing
How I quote.
I price the system against the value it creates for your team, not against another monthly subscription. One number is public; the rest are quoted in proposals, exactly, against scope.
True for everyone
- The Loop Diagnostic is $6–8k, fixed. Credited toward the build if you continue within ninety days.
- Module builds are fixed-fee against agreed scope. Most start around $10k, quoted exactly in the proposal. The stabilization window is included.
- Loop Care is a monthly retainer. Sized to what’s running in production and the cadence of new build.
If you want a number to plan against, I’ll give you a range in the first conversation. The proposal that follows will be exact.
FAQ
Common questions.
Can we take everything in-house later?
Yes. The work is yours from the first commit. If you bring it in-house, I support the transition: documentation, runbooks, access transfer. Some clients structure this from the start.
What happens if we stop working together?
You keep a running system either way. On managed infrastructure, you get a full handoff — source code, database export, runbooks, access transfer — and bring it in-house or hand it to a new operator. What stops is the tending: the tuning, the monthly session, and the memory layer that keeps learning from your corrections.
Do you work with our existing tools?
That’s the whole design. Shopify, Klaviyo, Triple Whale, Meta, your ERP, your inbox — each becomes an input to the system. I build between your tools, not on top of them.
Why is Loop Care part of the deal after a build?
Because software that nobody tends drifts, and a system that never hears your corrections never improves. Care covers everything that keeps it running — hosting, monitoring, upgrades — plus the feedback loop that makes month six sharper than month one. If your team runs its own infrastructure, Care is optional; the drift isn’t.
What’s the smallest engagement?
The diagnostic. Two weeks, fixed fee. Even if we never build, you leave with a map of your data, proof of what joins to what, a back-test against your own history, and a scoped proposal you could hand to anyone.
Do we have to book the diagnostic to start?
No. Start with a short call — twenty minutes, no fee — to see whether this is even the right fit. If it isn’t, I’ll tell you, and point you toward what is. The diagnostic is the first paid step, not the first step.
You’re one person. What if something happens to you?
Fair question — and it’s exactly why everything lives in your repository, not mine. If I’m ever unavailable, you still hold the source, the database, the runbooks, and the access. It’s documented as it’s built and it’s standard TypeScript, so any competent developer can pick it up — not a black box only I understand. That’s the difference between owning a system and renting access to one.
Rae Lakes Loop Trail, California
I’d rather prove it in two weeks than pitch it for two months.
The diagnostic exists so neither of us has to take anything on faith. Your data, your history, a fixed fee, a straight answer.