Sample · fictional company

This is what an AI Quickscan looks like.

This is a sample report for Testprospect B.V. That company does not exist. All findings, numbers and sources are made up to show the format.

Company: Testprospect B.V. (fictional)
Sources: made up, as a sample

In short

The core first, then the details.

Hypothesis

Enquiries through the website sometimes wait days for a reply. Writing up intake meetings costs the team hours every week.

To do this week

Count enquiries and response times for two weeks. Then a direct confirmation with a draft reply.

In the first month

A three-week trial with draft reports of intake meetings.

What we need from you

The number of enquiries and intake meetings per week.

Scorecard

Six areas, and how sure we are.

Choose an area for the explanation. Confidence runs from not visible to high. There is no score per area yet: that first needs fixed scoring rules.

Reachability for enquiriesForm without a confirmation of receipt. Confidence: HighOpportunity found

Source: Sample source C, the company website (fact).

What it means: A customer does not know whether the enquiry arrived.

Go to Q1

Response to customer questionsTwo of six reviews mention a slow reply. Confidence: MediumOpportunity found

Source: Sample source B, public reviews (observation).

What it means: Six reviews is not much. So count first, then solve.

Go to Q1

FindabilityNo FAQ and no structured data on the site. Confidence: HighOpportunity found

Source: Sample source C (fact).

What it means: Search engines and AI assistants cannot find the answers to common questions.

Go to Q2

Standard documentsThe intake report is a fixed step. How it is made is not visible. Confidence: LowOpportunity found

Source: Sample sources C and D (fact and assumption).

What it means: A vacancy for an office manager suggests pressure. That is an assumption.

Go to M1 Go to M2

Planning and follow-upNot visible in public sources. Confidence: Not visibleQuestion for the meeting

Source: No public source.

What it means: We ask this in the meeting. We do not fill it in ourselves.

Control over AINothing visible about AI use. Confidence: Not visibleQuestion for the meeting

Source: Assumption.

What it means: Who already uses AI, with which data? That is a question for the meeting.

Go to Q3

4 of 6 areas visible in public sources.

Opportunities

Six opportunities, from this week to 90 days.

Each opportunity has what we saw, a solution, the effort and a first step. A staff member approves every message.

Q1 · This week · Recommended first step

Count for two weeks, then confirm receipt

What we saw: Form without confirmation. Two reviews about waiting (confidence Medium).
Solution: First count enquiries and response times for two weeks. Then, if it matters, a direct confirmation with a draft reply. A staff member sends it.
Effort: Small
First step: Note the date and the time to reply for each enquiry.

Q2 · This week

FAQ page and findability

What we saw: No FAQ and no structured data (confidence High).
Solution: An FAQ from the ten common questions, readable for search engines and AI assistants.
Effort: Small
First step: Use the same ten questions as in Q1.

Q3 · Question for the meeting

AI agreements on one page

What we saw: Nothing visible about AI use (assumption, Low).
Solution: One page: which tools, which data never, who decides.
Effort: Small
First step: Ask who in the team already uses AI.

M1 · First month

Draft report from intake notes

What we saw: Intake as a fixed step. Office manager vacancy (confidence Medium).
Solution: AI makes a draft report in the company's own layout. An adviser checks it.
Effort: Medium, three-week trial with a decision point
First step: Collect ten notes from recent intakes.

M2 · First month

Standard documents as templates

What we saw: Question for the meeting: which documents do you make again every week?
Solution: Templates in the company's own style that AI fills in. A person checks.
Effort: Medium
First step: Choose the three documents made most often.

G1 · 90 days · Bigger step

AI employee Enquiry Follow-up

What we saw: Q1 and M1 touch the same work: the enquiry.
Solution: Combines Q1 and M1 into one fixed task with limits. Only if the trials work. See the card below.
Effort: Larger
First step: Decide after the review on day 28.

Impact versus effort

Top left is what comes first. Choose a point to see the opportunity.

Solid circle: this weekOpen circle: first monthSquare: 90 daysDotted: question for the meetingBordeaux: recommended first step

Impact and effort are an estimate on a scale of 1 to 5, not a measurement.

Plan

The first 30 days.

Week 1 the small steps, then trials with a decision point and a review.

Choose a day

Week 1: the small steps. Day 14: decision point. Day 28: review.

Q1 Count, then confirm

Q2 FAQ and findability

Q3 Question in the meeting

M2 Standard documents

M1 Intake report, trial

G1 AI employee, decide later

Day 1Day 15Day 30

Day 14: Decision point: counting is done, the M1 trial starts or not.
Day 28: Review and measurement. Then the decision on G1.

Worked example

Time saved, with your own numbers.

A worked example, not a promise. The saving is an assumption. In a real quickscan the company counts for two weeks first; only then is there a number.

Example: 10 enquiries times 15 minutes times 30 percent, minus 10 minutes of checking: about 0.6 hours per week.

Enquiries per week (your number) 10

Minutes per reply now (your number) 15 min

Saving (assumption) 30%

Checking per week (assumption) 10 min

Calculated result per week, not a measurement 0

No cost saving and no commitment. Whether time is freed up only shows in a measurement in practice.

Findability and safety

Being found, and starting safely.

Findability (fictional)
Check	Result
Index	Found with 14 pages (search engine, domain filter)
Name	A company with an almost identical name ranks higher
Assistant test, three questions, two providers	Mentioned for question a, not for b and c

Safety per step
Opportunity	Data	Who approves	Not yet
Q1	Name and question from the form	Staff member, every message	Sending automatically
Q2	No personal data	Owner, before publishing	Answers about individual customers
M1	Meeting notes	Adviser, every report	Special categories of personal data

Appendix

Card of an AI employee.

This is what a bigger step looks like before anything is built.

Card G1: Enquiry Follow-up

Problem: Enquiries sometimes wait days. Reports cost advisers time (assumption, to be confirmed).
What it does: Confirms every enquiry, summarises it and makes a draft report after the intake.
Trigger and input: Every new enquiry and every intake meeting. Input: form and notes.
Connects to: Mail and calendar: read and prepare drafts. Sends nothing.
Who decides: The staff member sends. The owner approves new standard answers.
Limits: No sending mail, no confirming appointments, no special categories of personal data.
When in doubt: Stops and flags the enquiry for a staff member.
Measure: Response time and minutes per report, with a baseline in weeks 1 and 2.
Privacy: Which data, where it is stored and a processing agreement: settled before use.

Sources and limits. Sample sources A to D are made up. In a real quickscan every finding has a source with a date and a confidence level. Findings with confidence Medium or Low are not confirmed. Nobody can guarantee findability in AI assistants.

What this sample shows

A quickscan is the first step of the approach: looking at where AI saves time, and where it does not. With one first step and a source for every finding.

Read the approach How JT Studio runs itself