AI readiness assessment: data scored, cloud checked, the build order priced.
A multi-company construction group deployed its first AI tool, then paused every new AI initiative for six weeks while a third party evaluated “our data capabilities and what we’re able to do” — and learned it doesn’t really have a cloud. Right question, backwards order. A readiness assessment asks it before the build.
Official services partner of the platforms defining AI
NVIDIA
Anthropic
The AI readiness assessment was supposed to come first.
From a check-in call with a multi-company construction group — an electrical contractor and its sister companies: one AI tool built and deployed, roughly thirty people waiting on rollout, and then every new AI initiative frozen for a six-week data-capability audit.
The pause came after the build
The tool was delivered — “the keys are sort of in your pocket” — and then came the freeze: six weeks of outside evaluation while all new AI work waits on the findings. The readiness question was the right one. It was just asked last.
“We don’t really have a cloud”
The infrastructure truth surfaced after the tool existed: no cloud environment to run it in, one engineer learning how to deploy on Azure, the team literally going back to the drawing board. Where AI runs is a readiness question — and nobody had asked it.
Data capability limitations, found late
The demo data behaved; the production data is another matter. What the company’s data capabilities actually are — which sources exist, what shape they’re in, what they can support — is being mapped for the first time, by an outside firm, after the money is spent.
The marketplace freeze
Every vendor conversation is on hold until the report lands — then they plan to “enter back into the marketplace on the tools.” Six weeks of momentum gone, and the re-entry still has no criteria beyond whatever the audit recommends.
Thirty people waiting on “ready”
Rollout is pending — around thirty users who need to be “ready to go when it comes time.” Nobody measured what ready means: no adoption baseline, no training plan, no definition anyone signed.
No build order, no prices
After the audit, then what? Which workflow is second, which is third, what each costs and what each pays back — the group has no roadmap for the next several weeks, let alone a priced order of work.
What skipping the AI readiness audit actually costs.
What it looks like when the readiness questions get asked after deployment instead of before the build. Your numbers will differ — the first two weeks of an engagement put figures on yours.
From an anonymized engagement — a multi-company construction group with one AI contract tool deployed and every new AI initiative paused
The cloud readiness assessment comes before the build order.
Two weeks inside your operation, a fixed fee, and an answer you own either way — including when the answer is no. Most AI implementation services pitches start at the build; this is the step that decides what the build should be.
- 01
Score the data capabilities
What your data can actually support: where contracts, prices, schedules, and customer records live, how current they are, what shape they’re in, and which candidate workflows they can feed. Data readiness for AI is the section vendor demos skim; here it is the first gate — a data readiness assessment in plain language, with the limitations named.
- 02
Check the cloud and tenant reality
Where would the system run — your Azure tenant, a tenant that doesn’t exist yet, something on-prem? Whose security perimeter, whose keys, whose logs? “We don’t really have a cloud” is a finding, not a failure: it prices the environment work into the plan instead of discovering it after deployment.
- 03
Price the build order
Each candidate workflow gets a payback estimate against build cost, assumptions marked. The output is a ranked order — what to build first, what second, what to leave alone — and, where the data or the environment can’t support a build yet, an honest no with the fixes priced.
Where an AI maturity model helps — and where it’s astrology.
The readiness market is two things: self-score quizzes from the platform vendors selling the destination, and AI maturity model frameworks from the AI strategy consulting firms. The vocabulary is useful — experimenting, scaling, embedded is a decent shorthand for where a company sits in a board meeting.
The astrology starts when a score out of five becomes the verdict. No maturity quadrant can tell you whether your contract data can feed an AI review tool, whether you have a tenant to run it in, or which workflow pays back first. Those things are measured, not scored — and that measurement is the assessment.
The assessment is also the before, not the after. Who owns the system once it’s adopted — the policy, the supervision, the ownership model — is AI governance consulting, a sibling engagement that starts where this one ends. And if your team is still in “we’re in discovery phase” mode, this is the same instinct with numbers attached.
We don’t really have a cloud — we’re literally going back to the drawing board.
Led by Peter Enestrom.
The AI readiness checklist, answered straight.
What operators ask before they let anyone score their readiness.
A structured answer to one question: is this company prepared to put AI into production — and if so, on what first? Ours scores three things: your data capabilities (which sources exist, what shape they’re in, what they can support), your cloud and tenant reality (where a system would run, whose perimeter, whose keys), and your workflow shortlist (which candidates pay back, in what order, at what price). Some buyers call the same engagement an AI readiness audit. Either way the deliverable is a written document you own — a ranked build order with the assumptions marked — not a score.
The working version: where your critical data lives and how current it is; whether a cloud tenant exists to run anything in; which workflows are high-volume and rule-based enough to automate first; what each would cost to build and to pay back; who would operate the tool once it ships; and what employees already do with AI tools on their own. If a checklist stops at strategy questions and never opens your data or your tenant, it’s a survey, not a readiness check.
Yes — and saying it out loud puts you ahead of the companies that find out after deployment. Readiness doesn’t require a cloud; it requires knowing what must exist before a build runs. Sometimes the answer is a tenant to stand up — that’s our Azure migration work, a sibling engagement — and sometimes the right answer is on-premise. Either way, the environment work gets priced into the plan instead of surfacing as a six-week pause.
A maturity model gives you vocabulary — experimenting, scaling, embedded — and that shorthand has its uses. What it can’t tell you is whether your data can feed the tool you want, whether you have somewhere to run it, or which workflow pays back first. A score out of five is astrology when it substitutes for measurement. The assessment measures; where the maturity talk is useful we use it, and where it isn’t we price the work instead.
No — readiness is the before, governance is the after. An AI readiness assessment decides whether and what to build. AI governance consulting starts after adoption: who owns the system, what the agents may do, who supervises them in production, and the usage policy around all of it. We run them as separate engagements in that order — assessment first, governance once there’s something in production to govern.
Then it paid for itself. A no comes with the reasons and the fixes priced — the data cleanup, the tenant standup, the sequence that makes the build viable — so “not yet” becomes a plan instead of a surprise. And the document is yours either way: if you never build with us, you still know more about your own operation than you did two weeks earlier.
Start with one workflow.
Tell us where your team loses hours. We will come back with a straight answer on whether AI can help, what it would take, and what it would pay.