The audit-sprint-retainer model: how to structure AI consulting engagements

When I was setting up Denver AI Tech, the obvious move was to do what every other AI consultancy does: a discovery call, a custom proposal, a six-figure scope. I built that structure on a whiteboard and threw it out the same day. Here is why, and what I built instead.

The proposal-Excel-sheet trap

The standard agency engagement looks like this: prospect contacts you, you do a 60-minute discovery call, you go away for a week and write a 20-page proposal with a custom-priced statement of work, you negotiate scope, you negotiate price, and four to six weeks later you start the actual project.

Three things are wrong with this for the mid-market.

First, the buyer cannot price-shop without doing this entire dance with multiple vendors. A COO at a 200-person logistics firm cannot put three of these proposals next to each other and choose. By the time they have three proposals they have invested twelve hours in calls, and the proposals will use different scope frames and different pricing models so they are not comparable anyway.

Second, the proposal becomes the artifact, not the work. Vendors compete on the elegance of the proposal. The proposal is rarely an honest forecast of the engagement; it is a sales document. After the contract is signed, the actual work bears only loose resemblance to the proposal.

Third, scope creep is the business model. The vendor is incentivized to write a proposal that gets the deal closed and then expand scope mid-engagement via change orders. A buyer who expects this and a vendor who profits from it are not aligned.

The mid-market buyer has watched this play out at every consulting firm they have hired. They want a different deal.

Productized pricing — and why three is the right number

Productized pricing means: published prices, fixed scope, fixed timeline, no statement-of-work negotiation. The buyer reads the page, sees the price, books a call only to confirm fit, and signs.

The objection most consultants raise is "but every engagement is different." Of course every engagement is different. So is every haircut, but every barber publishes a price for a haircut. Productized pricing does not mean every project is identical. It means the wrapper is identical, and the project inside fits the wrapper.

I picked three offers because mid-market AI engagements cluster into three distinct shapes:

The audit shape — the buyer knows AI could help but does not know where. They need a senior practitioner to look at their operations, identify the right opportunities, and rank them. This is two weeks of work and produces a written report. $1,500 standard, $750 founding partner.

The sprint shape — the buyer has a clear set of pain points and wants implementation. They are willing to commit to a fixed scope. This is four weeks and produces working software in their stack. $7,500 standard, $3,750 founding partner.

The retainer shape — the buyer has multiple ongoing operational improvements and wants senior support without a full-time hire. This is monthly with a 3-month minimum. $4,500 per month standard, $2,250 founding partner.

Three is enough to cover the realistic ways a mid-market firm starts an AI implementation engagement, and few enough that the buyer can read the page and self-route.

What productized pricing actually buys you

Three things, in this order:

1. Trust on the first call. A buyer who has read the price before the discovery call cannot be sold to. The dynamic shifts from "convince me to hire you" to "is this engagement the right fit." That is a much healthier conversation, and it produces better engagements.

2. Comparability across vendors. A buyer can put my $7,500 sprint next to a $35,000 fixed-scope engagement from a small consultancy and a $90,000 statement-of-work from a big firm. The decision becomes legible. Many buyers will pick a higher-priced vendor — that is fine. But they pick with full information.

3. Accountability on scope. When the price is fixed, both parties have an interest in keeping scope honest. Material scope changes get re-priced as a new sprint. The conversation is "should we do another sprint" not "let me invoice you for the extra work."

What it costs me

Two things.

The first is flexibility on weird engagements. Some genuinely unusual projects do not fit the three boxes. I have to either decline them or scope them as a new productized offer. So far the rate of weird engagements has been low enough that the trade is worth it.

The second is upside on big deals. A big consulting firm could land a $250,000 statement-of-work for a workstream where I am charging $7,500 plus a retainer. Some buyers will go to the bigger firm; that is fine, they were not my buyer to begin with. The mid-market buyer is not paying $250,000 for that workstream. They are paying $7,500 or nothing.

What buyers should look for

If you are evaluating AI engagements, here is what productized pricing should look like from the buyer side.

The price is on the page before you book a call. Not "starting at," not "depends on scope" — published.

The scope is itemized. Specific deliverables, specific timeline, specific success criteria. Not "AI strategy."

The vendor is willing to say no. If after the discovery call the engagement is not a good fit, they pass. A vendor who will take any engagement is selling, not consulting.

The structure is the same as the next buyer. You should be able to find another buyer who got the same offer at the same price. Productized means productized.

Pulling it together

The audit-sprint-retainer model is not the only way to structure AI consulting. But for mid-market buyers in 2026 — companies who have been burned by big-consulting proposals and are not ready for enterprise scope — it is the model that gets the work done with the least friction.

If you are building a consultancy, productize. If you are buying one, look for productized vendors. The proposal-Excel-sheet model is a relic, and the buyers who are tired of it are the buyers worth working with.