AI operations audit: what you actually get for $1,900

“AI audit” is one of those services that can mean anything, which is why most of them are worth nothing. Ours gets you three fixed things for $1,900: a written map of every automatable process we find in your business, a build list ranked by effort and ROI, and a 90-day roadmap saying what to build first. It ships in 10 working days, entirely in writing, with a full refund if nothing on the list is worth building.

Why most “AI audits” produce a deck, not a build list

A ranked, checkable list is rare because most AI engagements are built to survive contact with a slide, not a ledger. MIT’s NANDA initiative reviewed 300 enterprise generative AI deployments in its “State of AI in Business 2025” report (July 2025) and found 95% delivered no measurable profit-and-loss impact. Only 5% did.

Most of those engagements never produced anything a buyer could check. Gartner said in a June 2025 press release that it expects over 40% of agentic AI projects to be canceled by the end of 2027, citing escalating costs and unclear business value as the leading causes, not model quality. BCG’s September 2025 report “The Widening AI Value Gap” put a number on the same gap from the other side: 60% of companies it studied report minimal revenue or cost gains from AI investment, while only 5% qualify as “future-built” and generate substantial value at scale.

A vision deck cannot be checked against any of that. A ranked list with an effort score and an ROI estimate per item can be, which is the entire reason ours is built the way it is.

Deliverable one: the automation map

The automation map lists every process we can find in your written intake and granted access, each with three facts: what it costs you today in hours, errors, or delay, whether current AI can do it reliably, and what would have to be true for automating it to be safe.

It usually surfaces work the owner had stopped noticing they were paying for.

This is closer to an audit in the accounting sense than a brainstorm. We are not asking “where could AI help” in the abstract. We walk your actual workflow, the one running today with the tools you already have, and write down where time and money leak. Deloitte’s 2026 State of AI in the Enterprise report, based on a survey of 3,235 leaders run in August and September 2025, found only 30% of organizations were redesigning core processes around AI; the rest were bolting AI onto processes nobody had re-examined. The map exists so you are not one of that 70%.

Deliverable two: the ranked build list

The ranked build list scores every candidate automation on two measures: effort, meaning how long a competent builder needs, and ROI, meaning what the automation returns per month in hours or money. The list is then sorted by that pair, so the order of work is a number, not an opinion.

This sounds obvious. It is also the single output most AI consulting engagements never produce, because a ranked list is checkable and a vision statement is not.

We score your list with the same method we use to decide what our own agents build next, across the 41-agent fleet that runs our group. McKinsey’s Global Survey 2025 found that 88% of organizations report using AI somewhere in the business, but only about a third have scaled any of it past a pilot, and just 6% qualify as high performers capturing real EBIT impact. The gap between those numbers is a prioritization problem, not a technology problem: everyone can start a pilot, and almost nobody ranks what to build next by anything sturdier than enthusiasm.

Ranked items above a certain score become candidates for a tomte, our word for one production agent with one defined job, watched, owned, and logged like any other part of your operation. Items below the line get marked “skip” or “buy off the shelf,” in writing, so the list stays honest about what is not worth building too.

Deliverable three: the 90-day roadmap

The 90-day roadmap sequences the ranked build list into what to build first, second, and third, what each step should cost, and what to measure to know it worked. It is written to survive contact with any builder, ours or someone else’s; if you never speak to us again, it still tells you what to do next.

Sequencing matters more than the items themselves. Building the highest-ROI automation before the data feeding it is clean tends to produce a fast agent that answers confidently wrong questions. The roadmap orders around dependencies like that, not just raw score, and says plainly where a smaller fix has to land before a bigger one is safe to build. Ninety days proves the first two or three items and stays short enough that the plan does not go stale waiting on your calendar.

The refund clause, seriously

If our AI operations audit finds nothing worth building, the full $1,900 comes back. Some businesses genuinely should not automate yet, because their volumes are too low, their processes too unstable, or their data too scattered to trust an agent with it, and refunding that finding costs less than a padded report that burns your trust and our name.

The clause also answers the market numbers above. If nearly every AI engagement can claim “value delivered” regardless of outcome, as the MIT NANDA and BCG figures suggest happens broadly, the claim stops meaning anything. A refund tied to a specific written finding, “nothing here clears the bar,” is a claim you can hold us to. We would rather return the fee a few times a year than join the pile of engagements nobody can verify.

Audit vs. free consultation vs. full agency discovery

Businesses usually gather this kind of information through a free consultation, a full agency discovery phase, or a $1,900 AI operations audit. The three produce very different depth, and only one of them, the audit, is checkable against a number afterward, because it ranks every item in writing.

	Free consultation	Full agency discovery phase	Our $1,900 audit
Typical length	30 to 60 minutes	2 to 6 weeks, often billed hourly	10 working days, fixed price
Output	Verbal impressions, maybe a follow-up email	A discovery deck and a proposal	A written map, a ranked build list, a 90-day roadmap
Ranked by effort/ROI	No	Sometimes, informally	Every item, in writing
Refund if nothing’s worth building	Not applicable, nothing was paid	Rare; discovery fees are usually sunk	Full refund, in writing
What you can build from it alone	Little; it is a sales conversation	The proposal, if you sign it	Yes, with any builder, by design

A free consultation costs you nothing and returns roughly that much: a sales conversation shaped to get you to the next call. A full discovery phase at a software agency often runs weeks and several thousand dollars before you see a build list, and that list is usually written to justify the agency’s own proposal rather than to survive an outside check. The audit sits in the middle on price and ahead of both on what you can actually verify afterward, because every item carries a score instead of an adjective.

How it runs

The AI operations audit runs entirely in writing: a structured intake, questions answered within one business day, and a build list delivered on day 10, with no calls at any point. If you want the full scope, the FAQ, and the intake form, the audit service page has all of it.

If you want to check what a vendor should be asked before you pay for any audit, not just ours, the audit checklist covers the questions worth asking first. And if you want to see what the resulting build lists have turned into once they ship, the roundup of builds by industry shows the pattern across clinics, logistics, plantations, and a handful of others.

Start async: describe what should stop eating your week at /start. A human reads every submission and you get a written reply within one business day. No call gets scheduled, because none is needed to read a form.