Advanced · Team and organisation · module 119 of 155
From pilot to habit in your team
Introduce AI step by step: from a small trial with metrics to a shared way of working.
After this lesson
You plan a pilot with a well-defined task, a small group, a baseline measurement and an evaluation moment, and then make a well-founded decision to expand, adjust or stop.
In view · An improvement loop
From pilot to habit
One task, a small group and a baseline measurement before the start.
Short training, regular follow-up moments and open attention to doubts.
Compare with the baseline and choose: expand, adjust or stop.
Part 1 of 3 · Small trial
Your preference for less movement is on. Choose each part yourself.
Start small and measurable
A team does not change through one training course or a new subscription. For a pilot, choose one recurring task where AI is likely to help and whose result you can assess, for example draft answers to frequently asked questions. Work with a small group that wants to take part, with varying levels of experience. Before the start, record a baseline measurement: how much time does the task take now, how many errors occur and how satisfied are colleagues or customers (lessons 15 and 66)?
Learning, collecting questions and taking doubts seriously
Give participants a short training with examples from their own work, and a fixed place to share questions, good prompts and errors (lessons 10 and 54). Plan short check-in moments, for example fifteen minutes every week. Take doubts seriously: concerns about quality, privacy or job security are signals, not resistance. Discuss them openly and make clear what AI does and does not do in this task. Those who get the hang of it can help colleagues as a point of contact.
Evaluate and decide deliberately
After the agreed period, compare with your baseline measurement: total time including checking, quality, number of corrections and the participants’ experiences (lesson 94). Then decide explicitly: expand to more people or tasks, adjust and try again, or stop. Stopping is also a valid result if the gain does not materialise. If you expand, you record the way of working in your policy and a workflow sheet, and you keep measuring (lessons 70 and 100).
New practice example
From explanation to a usable result.
A pilot tests AI on one task with a small group, a baseline measurement and a decision point.
Fictional practice situationFictional: an accounting firm wants AI to help with standard letters to clients. Today a letter takes 12 minutes on average and 1 in 8 letters needs a correction.
View a possible solution
This is how it can be done
Four-week pilot with three staff members and five types of standard letter. Baseline measurement: 12 minutes and 1 in 8 corrections. Measurement points: total time per letter and number of corrections. A short review every week. After four weeks: expand if the time falls without more corrections, otherwise adjust or stop.
Why this works
The decision is linked in advance to measurable criteria. That way it does not depend on a feeling or on one striking example.
Make it slightly more difficult
Think of a doubt an experienced colleague might have about this pilot. How do you take it into account in the plan?
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
The decision is linked in advance to measurable criteria. That way it does not depend on a feeling or on one striking example.
A step further
Think of a doubt an experienced colleague might have about this pilot. How do you take it into account in the plan?
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 one well-defined task must precede Put together a small pilot group
- Choose one well-defined task must precede Record a baseline measurement
- Put together a small pilot group must precede Give a short training with your own examples
- Record a baseline measurement must precede Work together and collect questions and mistakes
- Give a short training with your own examples must precede Work together and collect questions and mistakes
- Work together and collect questions and mistakes must precede Compare with the baseline measurement
- Compare with the baseline measurement must precede Decide: expand, adjust or stop
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 plan a six-week pilot for a fictional customer service team of eight staff members. Task: draft answers to frequently asked questions with an approved AI tool. Give the pilot group, the baseline measurement, three metrics, the training, the weekly check-in moments, how we collect doubts and errors, and the criteria for expanding, adjusting or stopping after six weeks.
Apply it
After a one-hour training session, the whole team gets access to an AI tool. After two months, no one knows whether it achieved anything.
Which approach best fits this situation?
- Give another training session and then look again.
- Ask the tool itself how much time it has saved.
- Start again with a well-defined pilot: one task, a small group, a baseline measurement, measuring points and a decision moment.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
After a six-week pilot, the first draft is ready faster, but the total time per answer has stayed the same because of extra corrections. What do you do?
- Expand to the whole team immediately, because the draft is ready faster.
- Adjust, for example by improving the prompt or sources, and measure again; or stop if the gain does not materialise.
- Stop counting the corrections.
Extra practice
Choose one task in your team that is suitable for a pilot. Write down the baseline measurement, two metrics and a decision moment, and discuss the plan with a colleague who has doubts.
Task: draft answers to the twenty most frequently asked questions. Group: four volunteering staff members, both newcomers and experienced colleagues. Baseline measurement: on average 9 minutes per answer; 1 in 10 answers is corrected. Metrics: total time per answer including checking, and the number of corrections. Every Friday, fifteen minutes to discuss questions and errors. Decision after six weeks: expand if the time goes down without more corrections.
