Every AI consultancy will happily sell you a readiness assessment. Six weeks later you own a maturity matrix, a heat map, and a deck — and you are no closer to a working system. Readiness is not a score. It is a short list of conditions that determine whether a project ships or stalls.
After enough deployments, the pattern is clear: the companies that ship share five signals. None of them require a data science team.
01The five signals
- ▸A named owner. One person who wants the system to exist, has the authority to make decisions weekly, and will still be accountable in six months. Committees do not ship.
- ▸One measurable workflow. Not "improve operations" — a specific process with a countable unit: tickets resolved, candidates screened, invoices matched, renewals flagged.
- ▸Reachable data. The information the system needs exists digitally somewhere you can point to — a CRM, a ticket queue, a shared drive. Messy is fine. Missing is not.
- ▸Tolerance for a v1. The team accepts a first version that handles 70% of cases with a human catching the rest. If the bar is perfection on day one, the project dies in review.
- ▸A 30-day yes. Someone can approve the first deployment inside a month. If every decision routes through a quarterly steering committee, momentum dies before the build starts.
02The anti-signals
Just as predictive — in the wrong direction:
- ▸The goal is "an AI strategy" rather than a working system.
- ▸The sponsor wants a chatbot but cannot name the workflow it improves.
- ▸Success is undefined, or defined as "innovation."
- ▸The data lives in someone's head, or on paper.
- ▸Nobody is allowed to change the process the AI is supposed to improve.
03The readiness checklist
- ▢We can name the owner in one sentence.
- ▢We can name the workflow and its countable unit.
- ▢We know where the data lives and who controls access.
- ▢We have agreed what the human does when the system is unsure.
- ▢We know what number should move, and by when.
- ▢Someone can say yes to a v1 within 30 days.
OPERATOR NOTE — Score six for six and you are more ready than most enterprises with an AI council. Miss two or more and the fix is organizational, not technical.
04A worked example
Take a familiar candidate: renewal risk. The workflow is "catch at-risk accounts before the renewal date." Run the signals against it.
Owner: the VP of Customer Success — she has wanted this for two years and can decide weekly. Workflow unit: renewals flagged early, measured against the ones that churned unflagged. Data: CRM activity, support tickets, invoice history — all digital, all reachable. V1 tolerance: a weekly flagged-accounts list that is 70% right beats the current system, which is one person's intuition. The 30-day yes: the VP owns the budget line.
Five for five. That project ships. Compare it to "an AI assistant for the whole company" — no owner, no unit, no number, no v1 anyone can accept. Zero for five. That project becomes a deck.
05If you're not ready
Missing signals are fixable in weeks, not years. No owner? Appoint one — a person, not a working group. No measurable workflow? Shrink the ambition until it has a unit you can count. Unreachable data? The first project becomes the data path itself: wiring the workflow into a system of record. That is not a detour. That is the work.
Readiness is not a gate you wait behind. It is a checklist you clear — fast, deliberately, in order.
Ready to run it?
Put this manual to work.
A 30-minute strategy call with an operator — we'll map your first deployment path, not send a deck.