The pilot

Two weeks. One workflow. Real numbers.

The lowest-risk way to find out what private AI can do for your firm — proven on your actual documents, for a fixed fee, before you spend anything on hardware or deployment.

What the pilot includes

Week 0 — Scoping call & NDA

A 45-minute call to choose the one workflow that costs your firm the most hours. We sign an NDA before any confidential discussion, and agree the fixed fee in writing.

Week 1 — Build

You provide a small representative sample of documents under NDA. We build the private system around your workflow — on our lab hardware or a temporary machine placed in your office, your choice.

Week 2 — Evaluate

We measure the system against your real work: accuracy, time saved, failure cases, limitations. Your team tries it hands-on. Nothing is hidden — the failure cases are in the report too.

Delivery — The report & the decision

You receive a written evaluation report with all measurements, plus a fixed-price deployment proposal and exact hardware specification. All sample data is destroyed on a recorded session. If the numbers don't earn the next stage, it ends here — no lock-in, no pressure, and the report is yours to keep.

Pilot fee: from $4,500, fixed. The exact quote depends on the workflow and document types — that's what the form is for. Quotes are prepared by the engineer who would build your system, not a sales team, within 2 business days.

Not sure which workflow? Start with the Assessment (from $1,000) — we map your workflows, quantify the hours, and tell you honestly whether a pilot is worth it. Or just book a 15-minute call and talk it through first.

Request your pilot quote

No confidential details here — just enough to prepare a useful quote. NDA comes before everything else.

You'll hear back within 2 business days — from the engineer, with a fixed quote and a proposed scoping call time.

Before you ask

Three honest answers upfront

"Is the pilot fee refundable if it doesn't work?"

No — and here's why that's fair: you receive the full evaluation report either way, including the honest failure cases. A pilot that proves something doesn't work has saved you a $20,000+ mistake. That report has value in both directions.

"Can we skip the pilot and just deploy?"

We'll decline, politely. Deployment without measured proof is how AI projects fail and how vendors lose trust. The pilot is short and cheap precisely so that neither of us has to guess.

"Do we need to understand AI to do this?"

No. You need to know which task eats your team's hours — you already do. Explaining the rest in plain language is part of what you're paying for.