Density Labs / AI Readiness Assessment / What it includes

What an AI diagnostic actually includes.

"Diagnostic" and "assessment" get used loosely enough that the word tells you nothing about what arrives. Here is the actual contents, so you can compare it against anyone else's.

The five deliverables

DeliverableWhat it isWhy it matters
Workflow inventoryEvery candidate workflow we found, described in the language your team uses, with volume and who touches it.Most teams have eight to fifteen candidates and are arguing about two. The inventory usually reorders the list.
ROI and TCO rankingEach workflow ranked by return, with the running cost stated: inference, maintenance, and the human review that does not go away.Return alone picks flashy projects. Return net of total cost picks the ones that survive a year.
Data readiness checkFor the top candidates: where the data actually lives, who can grant access, how consistent it is, and what would have to change.This is the failure that kills pilots quietly. It is cheap to check and expensive to discover mid build.
90 day build planSequenced work for the top one to three, with the integration point named: which existing screen changes, and for whom.A plan that does not name the screen is a plan for a dashboard nobody opens.
A go or no-goA written recommendation, including do not build this where that is the honest answer.An assessment that always says build is a sales document, not a diagnostic.

How the two weeks are spent

What happensYour time
Days 1 to 3Interviews with the people doing the work, not only the people sponsoring it.Three to five conversations, 45 minutes each
Days 4 to 7Data access, sampling the real sources, and a technical read of the systems involved.One person to grant access and answer questions
Days 8 to 11Ranking, costing, and drafting the build plan for the top candidates.None
Days 12 to 14Written plan delivered and walked through live.One 90 minute session

What it is not

It is not a market scan, a maturity model, or a transformation narrative. It does not benchmark you against your industry. Those are strategy consulting deliverables and there are firms that do them well. This answers a narrower question: which specific thing should you build first, what will it return, and should you build it at all.

It is also not run by analysts. The person in the interviews is a senior engineer who would build the thing, which is why the plan comes with an integration point rather than a recommendation to identify integration points.

Who it is for

Mid market companies with one to three concrete workflows in mind and a real deadline. If you have fifty thousand employees and are rethinking your operating model, buy strategy consulting instead. If you have a customer support queue, a claims process, or a sales handoff that obviously wants automating and you need to know which one and whether it pays, this is the shape that fits.

Related reading: why 95% of AI pilots never reach production · in house vs an outside team · assessment vs strategy consulting