A sample AI readiness report
with every section marked Example.
This is the layout of the report you get after the assessment, filled in with an invented online shop. It shows how tasks are scored, how the top three are ranked and how we say where AI does not fit.
Invented company, invented data · Not a client result
Brands our teams have delivered for
This is a sample AI readiness report for an invented company, with every section marked Example. It shows the real layout. Tasks are scored on five checks, the top three use cases are ranked and one pilot is named. It also says where AI does not fit. It is not a client and not a result.
Everything below is invented.
The company, the people, the tasks and every finding were made up for this page. No real business, no real tool and no real result is shown. The report has the same sections as the one we write for you, so you can judge the method before you book. Your own report names your own tasks and carries your own evidence.
The summary and the one-line answer.
- Company
- Example Outfitters, an invented online shop selling outdoor gear, with about thirty staff.
- Scope
- Customer support, order operations and marketing. Three teams, six candidate tasks.
- Who took part
- Five people in two workshops: the owner, the support lead, the operations lead, the marketing lead and a finance clerk.
- The answer in one line
- AI fits three of the six tasks. Two should stay with people, and one is better solved with a fixed rule in a tool the shop already pays for.
- First pilot
- Drafting first replies to order questions, with a person reading and sending every reply.
What we looked at and what we found.
Every score later in the report points back to something in this list. If we could not see it, we did not score it.
| Source | What we reviewed | What we found |
|---|---|---|
| Workshop 1: support | How the three support staff handle order questions, step by step | Most questions repeat. Staff copy the same answers from old replies. |
| Workshop 2: operations and finance | How orders, returns and supplier invoices move between tools | Invoice checking is done by hand against a spreadsheet. Returns need a judgement call. |
| Help desk records | A sample of past tickets with names and addresses removed | Tickets fall into a short list of topics. A few ran to several replies. |
| Policy documents | The returns and delivery pages | The returns page is out of date after a courier change. |
| Product catalogue | An export of product names, sizes and supplier sheets | Sizes are written in more than one way. Supplier sheets are consistent. |
| Tools and access | The help desk, the order system and the accounting tool | The help desk allows a limited account. The accounting tool can match invoices by rule. |
The six tasks, scored on five checks.
Each task is scored on the five checks from the assessment page. High is good on every check. High on cost of a mistake means a mistake would be small and easy to fix.
| Task | Inputs | Volume | Checkability | Cost of a mistake | Fit with the team | Outcome |
|---|---|---|---|---|---|---|
| Draft first replies to order questions | High | High | High | High | High | Rank 1 |
| Sort the shared inbox by topic and urgency | Medium | High | High | High | High | Rank 2 |
| Write product descriptions from supplier sheets | Medium | Medium | High | Medium | Medium | Rank 3 |
| Match supplier invoices to orders | High | Medium | Medium | Low | Medium | Use a fixed rule |
| Approve refund requests | Medium | Medium | Low | Low | Low | Do not use AI |
| Plan next season's range | Low | Low | Low | Low | Medium | Do not use AI |
Our rule: a Low on checkability or on cost of a mistake moves a task to do not use AI, whatever else it scores. That is why refunds and range planning end at the bottom even though some of their other scores are fair.
The top three use cases, ranked.
The ranking weighs effort, expected value and risk together. Value is given in words, not figures, because the report does not promise a saving.
- 1
Draft first replies to order questions
The assistant reads the order question and the shop's returns and delivery pages, then drafts a reply for a person to read, change and send. It never sends anything itself. This is the most frequent task, and every draft can be checked in a minute against the policy pages.
- Effort
- Medium. Connect the help desk through a limited account and rewrite the out-of-date returns page first.
- Value
- High. It is the most repeated task and the first one the team named.
- Risk
- Low to medium. A draft might quote an old policy, so every draft cites its source page.
- Needs first
- A current returns page and a set of past questions with the right answers.
- 2
Sort the shared inbox by topic and urgency
Incoming messages are labelled by topic and urgency so the right person sees them first. A wrong label is visible straight away and a person can move the message with one click.
- Effort
- Low. The help desk already supports labels and the topics are a short list.
- Value
- Medium. It saves searching, not writing.
- Risk
- Low. A wrong label costs a few seconds.
- Needs first
- A list of topics agreed by the support team.
- 3
Write product descriptions from supplier sheets
A first draft of each description is built from the supplier sheet, in the shop's tone. A person reads every draft against the sheet before it is published, because an invented product detail is the main risk.
- Effort
- Medium. Sizes are written in more than one way and need one standard.
- Value
- Medium. It helps in busy weeks when new products arrive.
- Risk
- Medium. A wrong claim about a product reaches shoppers if no one checks.
- Needs first
- A style guide of a page and one agreed way to write sizes.
Where AI does not fit and what to do instead.
A good report says no as clearly as it says yes. These three tasks scored below the line, and each has a simpler answer.
- Approving refund requests: errors show up weeks later and cost money, so a person decides. A one-page rule card for the team makes the decisions faster and more even.
- Matching supplier invoices to orders: the accounting tool can match by a fixed rule, which is cheaper and easier to check than AI. Turn that setting on first.
- Planning next season's range: the knowledge sits in two people's heads and the task happens a few times a year. A shared sheet and a planning meeting would help more than any tool.
The first pilot and how you will know it worked.
The report ends with one pilot, a written way to judge it and a date to decide. If the case had not been there, this section would say so.
- Pilot
- Drafting first replies to order questions, for one queue, with the support team only.
- Owner
- The support lead, who decides and can say stop.
- Before it starts
- Time a sample of real replies for two weeks and count how many needed a second reply. This is the starting point.
- How it is judged
- The shop writes down what better looks like before the first test, using its own starting point. We do not set the target for it.
- Person in the loop
- A person reads and sends every reply. Nothing goes out automatically.
- Out of the pilot
- Refunds, changes to orders and any message sent without a person.
- Review
- A date in the diary where the owner and the support lead decide to stop, change or widen it.
Next step: if the shop wants to go ahead, we quote the pilot as a separate proposal. The assessment fee is credited against a build within 60 days, and the report is the shop's to keep whoever builds it.
What your own report
adds.
Your tasks, named by your people
The tasks come from two workshops with the people who do the work, not from a template.
Evidence behind each score
Every score points to a document, a record or something we saw, so you can challenge it.
Your documents and tools, read
We look at the real files and the real settings before we say what fits.
A pilot you may refuse
The report ends with one pilot and a way to judge it. You choose whether to go on, and the report is yours either way.
Where to go from a sample
Read the method, try the free tools, or book the week.
Frequently
asked.
No. The company, the people, the tasks and the findings are invented for this page, and every section says Example. We show no real client work here, so the page cannot be read as a result.
Because a report cannot know your numbers before it has seen your work, and we do not promise a saving. The sample gives value in words. Your report ends with a way to measure the pilot against your own starting point.
Yes, the sections are the same. They cover what we looked at, the five-check scores, the top three use cases, where AI does not fit and one pilot. The content is yours: your tasks, scores and findings come from your own workshops and files.
No. It ends with one pilot and a way to judge it, and it names where AI does not fit. The report is yours to keep, and anyone can use it. If you build with us within 60 days, the assessment fee is credited against the build.
Yes. Tick through the free AI readiness checklist in your browser. Or describe up to three repeated tasks in the free AI Automation Review, and a person sends a ranked shortlist.
Get a report like this
for your business.
Tell us which teams and processes to look at. We reply with a date for the assessment.
Fee credited against a build within 60 days · Reply within four business hours (typical)
Last updated
















