Services · priced in the open

Get implementation-ready. Then ship.

Five offers, ordered as a ladder. Start small: a fixed-price assessment or a rescue of the AI feature you already shipped. The deliverable at each step is scoped to be the plan for the next one — so you never buy a proposal, you buy work. Prices are on the page because your time matters; if a number is wrong for you, we’ve both saved a meeting.

Start with the assessment Or just email me

How the ladder works

Enter small.

A one-week assessment or a two-week rescue. Fixed price, fixed scope, useful on its own even if we never speak again.

The report is the proposal.

Every entry engagement ends with a scoped spec for the build — what to ship, the eval plan, the latency budget, the cost cap. No proposal-writing phase, ever.

Build, then keep it honest.

The Clarity Sprint or a dashboard pilot ships the thing. The ops retainer keeps the evals green and the costs flat after I hand it off.

Where to start

Entry · one week

AI Implementation Readiness Assessment

$5,000 fixed

Founding-client rate: the first three assessments run$2,500 in exchange for a named, written case study. When the third case study ships, so does this rate.

For teams that want an AI feature in a data-heavy product and need to know, before anyone writes code, whether the data, the evals, and the budget will hold.

Data-foundations audit — can your warehouse and dbt (or dbt-shaped) layer feed the feature you want, and what breaks first when it can’t?
Eval readiness score — what a golden set looks like for your use case, and what it takes to build one.
Latency and cost model for the feature, against your real traffic shape.
A decision-first feature spec, prioritized: what to ship first and what to skip.

The spec doubles as the Clarity Sprint scope. Roughly half of assessments should convert; the other half saved themselves a bad build.

Book an assessment Try the free cost estimator first

Entry · one to two weeks

Eval Rescue Sprint

$7,500 fixed

For teams that already shipped an AI feature and are watching it misbehave in production — without the harness that would say how, where, or how much it costs you.

50–100 golden examples sampled from your real traffic, labelled with your team.
An eval harness wired into your CI, so regressions block the merge instead of reaching users.
A cost monitor with an alert threshold you choose.
A findings memo: the failure modes, ranked by user impact, with the fix I’d ship first.

If the memo says “rebuild it,” the rescue price counts toward a Clarity Sprint.

Book a rescue Size your golden set free

The build

Flagship · two to three weeks

The Clarity Sprint

One AI feature, shipped in production, in your codebase. Eval-first: 50–100 golden examples before any production code, a latency budget and cost cap you sign off on, observability built in. The engagement flow is user moment → spec → eval setup → build → latency/cost validation → handoff. Boring is the goal.

Standard

$18,000 fixed

The feature, working, in your stack.
The eval harness and golden set, in your repo.
Cost monitor and latency validation against the signed budget.
A short doc: what shipped, why, and how to extend it.

With team enablement

$28,000 fixed

Everything in Standard.
Pairing sessions with your engineers throughout the build, not a demo at the end.
A second feature spec’d and eval-planned for your team to ship themselves.
A playbook for running evals and cost reviews without me.

Scoped by the assessment. If you skip the assessment, the first sprint week does that work — same flow, one bill.

Scope a sprint See shipped work

On your data, and after the handoff

Build · two to three weeks

Decision Dashboard Pilot

from $9,500 scoped fixed

For revops, sales ops, and fintech analytics teams: one decision surface on your warehouse, with an AI layer that narrates what changed and why — not a chatbot in the corner. The dashboards on this site are the pattern, rebuilt in the open; the pilot is that pattern on your data.

One dashboard answering one recurring decision — forecast trust, pipeline health, risk triage.
Automated insight narration with an eval set behind it, so the narration is measured, not vibes.
A governed dbt layer under the surface — tested models, documented lineage, and if the data is sensitive, a leak gate like the one on this site. The semantic definitions live in your repo, so the numbers survive me leaving.

Fixed price set after a scoping call — “from” covers a single-source warehouse; more sources, more scope.

Scope a pilot Browse the dashboards

Ongoing · monthly

AI Feature Ops

$3,000 / month

For teams I’ve built with. Every engagement hands off an eval harness and a cost monitor; this is me watching them so your team doesn’t have to.

Eval regression review on every prompt or model change your team ships.
Cost drift watch against the cap we set, with a monthly one-page readout.
Model migrations — when providers ship new models, I re-run the harness, migrate, and report the before/after. This happens more often than anyone budgets for.

Month to month, cancel anytime. Available after any build engagement above. Role Sprint clients get the wider Bench (same $3,000/mo, 10 hours of me included) — this, plus agent-run data refreshes and strategic on-call.

Ask about ops

Or hire the work, not the headcount

Role-priced · the 90-day build

The Role Sprint

$250/hr · the 90-day build $30,000

Founding-build rate: the first three builds run$15,000 — half the list price — in exchange for a named, written case study. When the third case study ships, so does this rate.

For any role I’ve already built as a JD case study — or the req you’re drafting now. Your posting prices 2,080 hours a year: base midpoint plus the ~25% employer burden, ramp and meetings included, and a new leader’s breakeven averages six months (Watkins, The First 90 Days). I sell the other side of that arithmetic: only the hours that ship systems.

The build — $30,000, ~120 hours, 90 days. The case study’s first-year plan shipped as working systems on your data. Compare it to the hire’s ramp alone, which typically costs three times that before the first deliverable.
The multiplier. Already hiring? Same build, handed to your new director on day one — their breakeven moves from month six-plus to month two. The req and I are not mutually exclusive.
The bench — $3,000/mo, 10 hours a month. The AI Feature Ops retainer widened for role work: 10 hours of me each month, evals green and costs flat, plus my agents running your data refreshes on a schedule (the same fleet that keeps this site current) and me on call. Month to month, cancel anytime — or prepay the quarter: $7,500 for three months, $1,500 off the month-to-month price, for paying up front.
The honest exit: if the audit says the role is three automations wearing a director’s title, I’ll say so, build the automations on the ladder above, and you skip the hire entirely.

Every JD case study on the site runs this arithmetic at the bottom of the page, from the range the company itself posted.

Email me the req Browse the case studies

Fit, stated plainly

Where I’m credible

Sales ops, revops, ad tech, analytics tooling, BI, and the modern data stack — fourteen years of it, and the features I build for clients are the ones I watch my own teams need. Decision-first specs, eval harnesses, latency budgets, cost caps.

Where I’m not

Model training, mobile apps, ML platform builds, and domains I haven’t operated in. If your problem lives outside my lane I’ll say so on the first call and, where I can, point you at someone better. That policy is cheaper for both of us than the alternative.

Not sure which rung is yours? Send two sentences about the decision your users are stuck on and what you’ve shipped so far. I’ll reply with the offer I’d pick — or tell you that you don’t need me yet.

Email me Or follow the work first