Skip to content

Rates & fit check

Published prices for making your AI costs legible

Productized engagements for teams whose AI and cloud bills grow faster than their usage, and for teams who want to ship more with AI-assisted delivery without losing control of quality or cost.

Start the 2-minute fit check

Cash-pay · advisory from $250/hr

  • Advisory hours

    A senior second opinion on an AI cost, architecture, or AI-assisted delivery decision before you commit to it.

    $250/hr

    2-hour minimum

    • Live working session
    • Written notes with the decision and its tradeoffs
    • Follow-up questions by email for a week

  • Executive Cost Briefing

    A board-ready story on your AI infrastructure economics: what it costs, who owns it, where the margin risk is.

    $2,500

    90 minutes

    • Cost & attribution landscape review
    • Unit-economics narrative
    • Board Q&A prep

  • Unit Economics Sprint

    Most common start

    A focused teardown of one AI system or workload: what it really costs per unit, and the top three levers.

    from $5,000

    1 day

    • Fully loaded cost per request, customer, or output
    • Spend breakdown: inference, compute, storage, egress
    • Top 3 levers, quantified
    • One-page findings

  • AI Cost & Unit Economics Audit

    The full teardown: attribute the spend to owners, model unit costs, and rank the waste by dollar impact.

    $12k–$25k

    1–2 weeks

    • Cost-attribution map (who owns what)
    • Unit-economics model
    • Waste & rightsizing findings ranked by $
    • Showback model finance can run

  • ML Cost & Reliability Evals

    An eval harness that treats cost as a first-class metric next to quality and latency, so cost regressions get caught like bugs.

    $15k–$30k

    2 weeks

    • Cost-per-inference baselines & budgets
    • Eval harness with cost gates
    • Drift & spend monitoring
    • Team handoff

  • AI-native rebuild

    Rebuild a website or internal tool with AI-assisted delivery I manage and review: smaller codebase, cheaper to run, verified before launch.

    Fixed quote

    After a Sprint

    • Scoped plan with cost-to-run estimate
    • Build behind human review and verification gates
    • Deploy pipeline & runbook

How it works

From form to findings in four steps

  1. 1

    Fit check

    Two minutes: your situation, budget, and timeline. If it’s a fit, my direct line appears at the end of the form.

  2. 2

    First call

    Thirty minutes, free. We pick the unit that matters and decide whether a Sprint, an Audit, or advisory hours fits.

  3. 3

    Fixed scope, one invoice

    A written scope with a price. Cash-pay by ACH or card, invoiced directly by me.

  4. 4

    Delivery with receipts

    Findings tied to real numbers from your systems, the levers ranked by dollars, and a handoff your team can run.

Fit check

Two minutes. No sales call required.

Tell me what’s going on and what you’ve budgeted. If it’s a fit, you’ll see my number before you submit. If not, I’ll point you somewhere useful.

The form runs on TendForm, a product I built and run solo. Here’s how I kept it production quality.

Open in a new tab ↗

FAQ

Questions people ask first

Something else? Put it in the fit check’s “what’s going on” box.

What does “cash-pay” mean here?

I invoice you directly and you pay by ACH or card. I don’t work through staffing platforms, marketplaces, or agencies, which is part of why the prices are published.

What is an “AI-native rebuild”?

Replacing a website or internal tool with a smaller system built mostly by AI agents, behind human review gates. The goal is lower run cost and less code to maintain. It’s quoted after a Sprint, once we know what the current system costs.

Will agents touch our code or data?

Only if you want delivery work, and only inside the guardrails we agree on up front: scoped repository access, allowlisted commands, no production credentials or spend in the agent’s toolbelt, and every change reviewed before it ships.

Will you bring anything from your big tech roles?

The judgment and the patterns, yes. Anything confidential, never. I don’t take work that conflicts with prior commitments, and every engagement starts from your systems and your numbers.

We don’t know our AI spend at all. Too early?

No, that’s the most common starting point. The Unit Economics Sprint exists for exactly that: one system, one day, a defensible number.