Advanced · AI in your sector · module 135 of 155
Project lesson: an AI application for your sector
Design an AI application for your own sector, with the rules, data and checks that apply there.
After this lesson
You choose a task from your sector, set out sector rules and data agreements, test with fictional cases and decide with a pilot.
In view · How it connects
An AI application tailored to your sector
A recurring task, with who uses the result and what goes wrong if there is an error.
Professional rules, the GDPR and a traffic light for what is allowed.
A test set with fictional cases, a pilot and a method card.
Part 1 of 3 · Task and consequences
Your preference for less movement is on. Choose each part yourself.
Choose a task that fits your sector
Every sector has its own tasks, rules and sensitivities: care works with health data, education with minors, customer service with customers’ expectations, a small business with prices and terms (lessons 131 to 134). Choose a recurring task whose result you can assess and where an error does not have serious consequences. Describe who uses the result and what goes wrong if it is incorrect.
Set out the rules and data
Find out which rules apply in your sector: professional rules, sector agreements, the GDPR and possibly the AI Act (lessons 116 and 117). Use a traffic light to decide which data may go into the application (lesson 118). Choose a tool your organisation approves. Write a test set with ordinary cases and edge cases from your sector, with fictional data (lesson 113). Record who checks the result and when a person takes over.
Test, pilot and method card
Run the test set and improve until the important cases pass. Start a small pilot with a baseline measurement and a decision point (lesson 119). Save the method in a method card: task, source, prompt, checks, sector rules and second route (lesson 100). Discuss the result with colleagues or a professional association: they often see risks that are specific to your profession.
New practice example
From explanation to a usable result.
A good AI application fits the rules, data and risks of your sector.
Fictional practice situationFictional: an estate agency wants to use AI to write property listings based on the property’s fact sheet, with the EPC label and the photos.
View a possible solution
This is how it can be done
Source: only the approved fact sheet, with floor area, EPC label and facilities. Limit: do not add features that are not in the fact sheet, such as “quiet neighbourhood” or “ready to move into”. Check: the estate agent compares every listing with the fact sheet before publication. Test case: a property without an EPC label in the fact sheet, where the listing must not invent a label.
Why this works
A listing with a wrong feature can mislead a buyer. The fact sheet as the only source and a fixed check limit that risk.
Make it slightly more difficult
Which rule from your sector would you include first in the limits of your prompt?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
A listing with a wrong feature can mislead a buyer. The fact sheet as the only source and a fixed check limit that risk.
A step further
Which rule from your sector would you include first in the limits of your prompt?
Try & discover · no obligation
Put the steps in a workable order
Move the steps with the arrows. Check whether each step has its necessary input and control.
What does a step depend on?
- Choose a task from your sector must precede Find the rules that apply in your sector
- Find the rules that apply in your sector must precede Use a traffic light to set down which data are allowed
- Use a traffic light to set down which data are allowed must precede Write test cases with fictional data
- Write test cases with fictional data must precede Start a small pilot with a baseline measurement
- Start a small pilot with a baseline measurement must precede Have a peer in your field assess the result
- Have a peer in your field assess the result must precede Save the working method in a method card
Move a step and then check the sequence. The arrows also work with your keyboard.
This model checks dependencies, not duration, capacity, or actual execution. With multiple agents, the human decision remains a separate step.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Help me design an AI application for this fictional task in my sector: [describe your sector and task]. First ask me about the rules in my sector, the data involved and who uses the result. Then make a plan with a traffic light for data, checkpoints, five test cases and a four-week pilot. Do not give legal advice; mark what I need to check.
Apply it
You are designing an AI application for an association that organises youth camps. The plan uses the registration forms with medical sheets to create the participant lists automatically.
Which approach best fits this situation?
- Keep medical data out of the application and create participant lists with only the necessary data; the medical sheets stay in the secure system with the person responsible.
- Use everything; it’s handy to have all the information in one place.
- Use the medical sheets, but put them in a shared folder so all the leaders can see them.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
Why do you have a plan for an AI application reviewed by someone in the same profession?
- Because AI tools work differently in each sector.
- Because people in the same profession see risks and rules that are specific to that profession.
- Because a plan is only valid with two signatures.
Extra practice
Work out the plan for a task in your own sector. Have it reviewed by a colleague in the same profession and note which risk they saw that you missed.
Sector: childcare. Task: drafting the weekly message for parents. Rules: no names or photos of children without consent, no information about health. Traffic light: green for activities and menu, red for children’s data. Check: the person in charge proofreads every message. Test cases: a week with an outing, a week with a carer off sick. Pilot: four weeks, with time per message as the measure. A colleague pointed out that even a description of an activity can make a child recognisable.
