An AI readiness assessment should end with things you can act on: the value at stake in each workflow, your candidates in order, and a list of what you will not build. Maturity scores rarely tell an owner what to do next. This guide walks through the method, so you can run it yourself or have StrikeArc Scout run it for you.
What an AI readiness assessment should produce.
Most companies start AI from the wrong end: a tool someone saw at a conference, a vendor demo, or a board member asking for the plan. An assessment starts from the work. Done well, it hands the exec who owns the problem a short document they can defend to a CFO.
- The value at stake, workflow by workflow. Each one tagged by what it moves: REV (revenue up), COST (cost down) or TIME (time back).
- A ranked shortlist. The candidates in the order you should take them.
- The do-not-build list. The ideas you are setting aside, and why.
If an assessment ends with a maturity score and no ranked list, it has left the owner's real question unanswered.
Step 1: list the workflows where time and money go.
Start with work that repeats every day or week. For each workflow, write down who does it, how often, and how long it takes.
In a traditional business.
Slow quoting, missed customer inquiries, manual order entry, invoice processing, dispatch coordination and the report someone rebuilds by hand every week.
In a B2B software company.
Lead qualification and follow-up, customer onboarding, support triage, CRM updates, billing coordination and repetitive reporting.
Where to look.
Shared inboxes, the spreadsheets that feed the ERP, anything your team types twice, and the report that takes Friday afternoon.
Step 2: score each workflow.
Score each candidate from one to five on five tests. The workflows that score high on all five are where AI pays first.
| Test | Scores 1 when | Scores 5 when |
|---|---|---|
| Volume | It happens a few times a month. | It runs daily, many times over. |
| Cost of delay | Nobody notices a slow day. | A slow answer loses a job, a discount or a customer. |
| Cost of errors | A mistake is caught and fixed in minutes. | Errors reach customers, invoices or cash. |
| System access | The work lives in people's heads or on paper. | The inputs arrive by email or live in the ERP, the CRM or shared files. |
| A named owner | Nobody owns the result. | One executive owns it and can approve changes. |
The full scoring sheet, with the processes to leave alone, is in the guide to what business processes can be automated.
Step 3: check system access and data.
Workflow AI needs read access to the systems where the work happens, real examples of the documents it will handle, and a person who can answer questions about the edge cases.
Read-only connections come first. A read-only connection lets a system see what your team sees without changing anything, and you can revoke it at any time. Write access, where a play posts a bill or updates a record, comes later and only where your approval rules allow it.
The bar for "clean enough" is lower than most owners expect. Analytics needs years of tidy history. Workflow AI needs this month's inbox, your price book and the records a person already uses to do the job.
Step 4: put numbers on the top candidates.
For each of your top three, work out what a working system returns in a year. The math has one part for each lever.
- REV. Revenue that stops slipping away: quotes sent the same day, inquiries answered before a competitor calls back.
- COST. Dollars that stop leaking: duplicate payments, billing errors, late fees, AI and software spend nobody uses.
- TIME. Hours given back, multiplied by the loaded cost of the people who do the work today.
Replace each input with your own. Before late fees and missed early-pay discounts.
Subtract what the system costs to run each month and you have the number to decide on. Leave out anything you cannot measure today; the readout should hold up in front of your CFO.
Step 5: decide, and write the do-not-build list.
Pick the first workflow, name its owner and set the measure you will judge it by. Then write down what you will not build, and why. The list protects the budget and your team's attention, and it stops the same idea coming back every quarter. Good candidates for it:
- Tasks that happen a few times a year.
- Work that leaves no data trail.
- Decisions that need a license or a signature.
- Processes that are about to change anyway.
- Work that nobody owns.
Where AI tends to pay first.
Sales.
Inquiries that wait for a callback, quotes that wait on one person, and follow-up that slips when the week gets busy. The guide to automating quotes and follow-up covers each one.
Back office.
Orders keyed in by hand, vendor invoices matched line by line, and reports rebuilt every week. Start with the guide to back-office automation, or go straight to invoice processing automation.
Service and support.
Missed calls, repeat questions and answers that live in one person's head. Several plays in the StrikeArc PowerPlays Library start here.
Finance.
Duplicate payments, unclaimed deductions, and margin or cash problems that show up a month late.
For a worked example in one industry, read AI use cases in distribution.
A readiness checklist you can use this week.
Answer yes or no. If you can answer most of these for a workflow, it is ready for a real system.
- Can you name the three workflows your team complains about most?
- Does each one have a single owner?
- Can someone export a month of real examples?
- Is there a system of record for each?
- Do you know today's cost or delay for each?
- Do you know who approves access to the systems?
- Do you know what you would refuse to automate?
- Do you know what budget range is realistic this year?
From assessment to a first-year plan.
Take one workflow first. Install it, run it for a quarter, look at the numbers, then decide on the next one. Each step asks the same four things of the business: an owner, access to the systems, real examples, and a baseline to measure against. Here is how find, install and run work with us.
StrikeArc Scout is the free diagnostic: an intake questionnaire and read-only connections to the systems you choose. StrikeArc Scout scores every play in the StrikeArc PowerPlays Library against your answers, and a person walks you through the readout: the value at stake, the top plays and the do-not-build list. The readout is yours to keep either way. If you are weighing paid options, here is what AI help costs.
The topics in this guide.
Questions owners ask.
What is an AI readiness assessment?
A structured review of which workflows in your business AI can take over. It ranks them by value and names what not to build.
What does an AI readiness assessment cost?
StrikeArc Scout is free. Paid assessments vary by provider; the guide to what AI consulting costs explains how they are priced.
Is my data ready for AI?
For workflow AI, readiness means access to the systems where the work happens and real examples of the documents. A perfect data warehouse is not required.
What should we automate first?
A high-volume workflow with a clear cost of delay or error, readable systems and one owner. The scoring method above ranks them.
Do we need an AI strategy before an assessment?
The assessment produces the useful part of one: a ranked list and a do-not-build list.
What happens after the assessment?
You decide. Install the top candidate, wait, or do nothing. The readout stays yours.