What is
AI readiness?
One job, ready
for one tool.
AI readiness means one named job in your company is ready to hand to an AI tool. Someone checks what comes back, and in 90 days you know whether it paid off.
The request usually comes with a deadline. Your operations lead has seen a demo that drafts proposals in under a minute. The trial ends Friday, and they need a yes.
The demo worked. It can’t show you how the tool does on your work.
The demo
- Runs on the seller’s own examples, cleaned up for the purpose.
- Drafts a proposal in under a minute.
Your work
- Runs on a job nobody has written down, and a folder holding three versions of the price list.
- Needs one person to read every draft before any client gets it.
AI readiness is about one job, not the whole company
You aren’t judging the whole company’s readiness for AI. You’re deciding whether to hand this one job to this one tool, this quarter.
A company can be ready for one job and nowhere near ready for the one beside it. Five checks settle it.
Count the yeses: five approves the spend, three or four approves one small step, two or fewer means fix first.
Search the term and you’ll find readiness indexes built for large enterprises. You’ll also find setup wizards from software companies, and tools for schools and governments. Many score you on five pillars, usually strategy, data, technology, people and governance.
That’s common vendor shorthand, not a standard. It answers a much bigger question than the one on your desk.
Run five checks before you approve the spend
Each check has one person in the company who can answer it. If anyone has to guess, count the answer as a no.
A no doesn’t judge the company. It names the first job to do, and none of the five fixes means buying anything.
Give the job an owner and a number before the trial ends
The job’s owner is whoever’s week changes if the tool works.
Often that isn’t whoever found the tool. They own the enthusiasm. They won’t be the one fixing drafts at 4 p.m.
“Help with sales.”
A department.
First drafts of every proposal.
A job.
The number is the other half. Pick the count that should change, and take it by hand before anything is installed. The AI consultant guide walks through that count. The method works the same way here.
If nobody will put a name on the job, don’t approve the tool.
Write the job in one line with the owner’s name beside it, before Friday.
Find the information before the tool needs it
An AI tool works from what you hand it. If the current price list lives in three places and one is out of date, the tool picks one. It won’t tell you which.
Here’s the test: have whoever does the job gather everything they’d need for one real example, from scratch. Watch the clock.
A few minutes.
Everything comes from one place.
An afternoon.
A call to someone who left last year, or the words “I just know.”
A no here makes the first project a tidy-up. One folder, one current version of each document and a name on it. That costs less than any tool, and the tool would need it anyway.
Make sure someone on staff can check what comes back
The check is a person. Name who reads a draft before it reaches a client. Then ask them one question: could you spot a wrong one?
If their honest answer is “I’d have to redo it to know,” the tool saves nobody time. It shifts the work to a later step, where mistakes are harder to catch.
Write down what may go into the tool before anyone signs up
A paste rule is one line on what staff may put into an AI tool. It also names what never goes in:
Write it before the trial starts. A trial is when people paste real work in to see what happens. So the rule has to exist on day one.
The rule itself is short. The guide to ChatGPT and confidential information has the version I’d start from: paste the problem, not the person. For readiness, what counts is that it’s written down, someone has signed it and the team has seen it.
With three or four yeses, approve one small step
Five yeses.
Approve it, and put the day you’ll compare the number on the calendar.
Three or four.
Approve one small step, small enough that your team could pick the work up again by hand. Fix the noes while it runs.
Two or fewer.
Spend the next month on the noes. None of them needs software.
Your own count, not the demo, decides the next one.
Questions owners ask about AI readiness
What is AI readiness?
Whether one specific job is ready to give to an AI tool this quarter. You settle it with five checks, and no fix on the list requires software.
What are the five pillars of AI readiness?
There’s no agreed set. Most vendor lists name some mix of data, people, technology, governance and strategy. For one spending decision, five practical checks on a single job tell you more.
How do I know if my company is ready for AI?
Pick the job you’d hand over and answer five yes-or-no questions about it. Five yeses means go ahead. Three or four means try it small first.
What is an AI readiness assessment?
A questionnaire that rates how prepared a company is to use AI. Many are built by software sellers for large enterprises, so check who built one before you trust its score. We built one too, the free AI check, and it gives no single score.
What should we fix first if we’re not ready for AI?
The checks that came back no. None of the fixes requires buying software. You need an owner, one folder, a checker, a written paste rule and a count.
How long before we know whether an AI tool paid off?
Set the date when you approve it, and compare against the count you took by hand beforehand. Without that count you can’t tell either way.
Your first fix costs less than the tool.
Answer 20 questions about how your business runs, and get the one gap in your way named, with fixes you can make this week.
Read next
Hiring help
What does an AI consultant do? Decide what you’ll own when they leave, before you sign.
October 4, 2026
Data safety
Is ChatGPT safe for confidential information? Not in an account you don’t own.
August 6, 2026
Policy template
AI policy template? Eight lines, and one name that approves every new tool.
August 3, 2026
