How do you move AI out of pilots and into daily work, in writing?
Many AI pilots never reach daily use. S&P Global reported in March 2025 that companies scrapped 46% of their AI proofs of concept on average before production, and that 42% had abandoned most of their AI initiatives. Gartner’s July 2024 prediction on abandoned generative AI projects pointed to poor data quality, inadequate risk controls, rising costs and unclear business value.
This workshop writes down how your organisation moves use cases into daily work. In one day, a cross-functional group selects and scores use cases and sets the briefing and verification methods, the data rules and the governance. Each participant then makes a first build of one use case, tested against its own test set. The playbook that records all of it is the organisation’s own document.
After this course
- Can you select and score AI use cases on value, data, risk and how easy the output is to check?
- Can you write briefs that give an AI tool the whole job in the first message?
- Can you set verification methods that check AI output against a second source or a person?
- Can you build and test a first assistant or automation, and know where a quick prototype stops being safe?
- Can you apply data rules, the AI Act and GDPR to each use case in the playbook?
- Can you set measures for adoption, quality and cost for each use case before launch?
Skills you practise
- Use-case scoring
- Briefing methods
- Verification
- Assistant building
- Automation testing
- Data rules
- AI governance
- Adoption measurement
Course structure
The workshop is 1 day of 6 teaching hours, in 2 blocks of 3 hours with a lunch break. Online, it runs as 2 live sessions of 3 hours on the same day, not recorded.
The morning covers selection and method. The group collects candidate use cases from each function, scores them and sets the data rules for the top items. Briefing and verification methods follow, practised on backlog items.
The afternoon starts with the build. Each participant builds a first assistant or automation for one use case in your approved tools, with the test set written before the build. It also covers vibe coding, where the AI writes code from a plain-language description. That is quick for a prototype, and the result stays away from live data, logins and permissions until someone has reviewed it. The UK National Cyber Security Centre said in December 2025 that prompt injection may never be totally mitigated, so each build gets only the access its task needs.
Governance and the legal rules come next: the AI use policy, the approved tool list, decision rights, the AI Act with its high-risk dates and GDPR. The day ends with adoption measures and the playbook draft.
About two thirds of the time is hands-on.
1Selecting and scoring use cases75 min
- Collecting candidates from each function
- Scoring on value, data, risk and ease of checking
- Why pilots get abandoned
- The scored backlog
A demo shows a tool on examples the vendor chose. Score each use case after testing it on examples from your own work, including the awkward ones.
Caution. Gartner predicted in July 2024 that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025. It named data quality, risk controls, costs and unclear value. Score those 4 before you build.
You leave with: A scored use-case backlog for the organisation.
2Data rules for each use case45 min
- Data classes and which tool may take each
- Personal and special-category data
- Processor contracts and training on inputs
- Reporting when data goes to the wrong tool
A tool that reads files, mail or web pages can be redirected by instructions hidden in them and leak what it can reach. The data rule for a use case covers what the tool can reach as well as what staff paste in.
Caution. Under GDPR, an unauthorised upload of personal data to an AI tool can be a personal data breach, notifiable within 72 hours where it poses a risk.
You leave with: A data sheet for each backlog item: fields, data class and approved tool.
3Briefing and verification methods60 min
- All requirements in the first message
- Reusable brief templates
- Checking each output in small steps
- Checking against a second source
A model answers from the text in the current chat, and answers get worse when the requirements arrive over many turns. Put the whole brief in the first message, and start a new chat when the task changes.
Caution. In June 2023 a US court fined 2 lawyers who filed 6 cases that ChatGPT had made up. One of them had asked ChatGPT whether a case was real, and it said yes.
You leave with: A brief template and a verification method for the top backlog items.
4Building and testing a first assistant75 min
- Assistants with instructions and reference files
- Simple automations in approved tools
- Vibe coding: fast prototypes and their limits
- Test sets and failure logs
- Hidden instructions in files and emails
An assistant or prototype built with AI can work on the first try and still fail on cases nobody tested. Write the test set first, and keep anything built this way away from live data until someone has reviewed it.
Caution. In June 2025 a flaw named EchoLeak showed that an email with hidden instructions could make Microsoft 365 Copilot send internal data to an attacker. Give assistants access only to what the task needs.
You leave with: A first build of the participant's use case, with a test set and a failure log.
5Governance, policy and the legal floor60 min
- The AI use policy and approved tool list
- Decision rights for new use cases
- AI Act: Article 4, Article 50 and high-risk dates
- GDPR: Articles 5, 9, 22 and 28
- Records for each use case
The playbook states for every use case whether the AI suggests, drafts or acts, and who approves the result. That level changes only through the decision rights the playbook sets.
Caution. Since 2 August 2026 AI Act Article 50 has required AI systems that interact with people to be designed so people know they are dealing with AI, unless that is obvious.
You leave with: The playbook's rules section: policy, tool list, decision rights and the legal check for each use case.
6Adoption, measurement and the playbook draft45 min
- Owners and users for each use case
- Baselines and 3 to 5 measures
- Quality next to cost
- Assembling the playbook
Each use case in the playbook names its owner, the people whose work it supports and the measure that shows it helps. A use case without an owner does not go live.
Caution. In May 2025 Klarna's chief executive said its AI cost-cutting push in customer service had gone too far, so measure quality and customer outcomes alongside cost.
You leave with: A first draft of the AI Playbook, with measures and baselines for each use case.
Selecting and scoring use cases
Give the criteria and weights, and ask for a reason per score.
Give me some AI use cases for our company.
Score the 12 candidate use cases below from 1 to 5 on: value (hours or errors saved per month), data sensitivity (5 = least sensitive), risk if the output is wrong (5 = lowest risk), and how quickly a person can check the output. Weights: 30, 20, 30, 20. Give a table with each score, a one-line reason and the weighted total, sorted highest first. Flag any score where you had to guess. [candidates]
Data rules for each use case
Classify each data field before it enters the tool.
Can I use customer data in this automation?
Below are the 14 data fields our support automation would use. Label each field public, internal or personal, and mark special-category data separately. For each field, say which of our approved tools may receive it, using the tool list below. Table format. [fields] [approved tool list]
Briefing and verification methods
Name every output, the order and the limits in the first message.
Summarise these supplier contracts.
For each of the 6 supplier contracts below, extract the notice period, renewal date, liability cap and governing law. Output one table, one row per contract, in the order given. Where a term is missing, write 'not stated'. Give the clause number for each value. Under 200 words outside the table. [contracts]
Building and testing a first assistant
Write the test cases before you build the assistant.
Make me a chatbot that answers HR questions.
Before we build an HR policy assistant, write 15 test questions: 8 it should answer from the attached policy, 4 it should refer to HR, and 3 that try to make it reveal personal data. For each, give the expected answer or action. Table format. [policy document]
Governance, policy and the legal floor
Paste the article text and ask which use cases it touches.
Do the AI laws apply to our use case?
Below are our 5 top backlog items and the text of AI Act Article 50(1) and 50(4) and GDPR Article 22. For each item, say which article could apply, quote the words that trigger it, and list what we must do. Mark 'ask counsel' where the text is unclear. Table format. [backlog items] [article text]
Adoption, measurement and the playbook draft
Ask for measures with a baseline, a source and the decision they inform.
How do we measure AI ROI?
For the use case below, propose 4 measures: one for time, one for quality, one for user adoption and one for risk. For each, give the baseline we must record before launch, the data source, the target after 8 weeks, and the decision it informs. Table format, under 200 words outside the table. [use case summary]
The organisation leaves with a first draft of its AI Playbook. It holds a scored use-case backlog, the briefing and verification methods, the data rules and governance, measures for each use case, and a first tested build per participant.
In the weeks after the workshop, each use case owner finishes the build, runs the test set again and records the baselines before launch. The next items on the backlog get owners.
The workshop has no exam and no grade, and it leads to no formal qualification, professional designation or recognised award. Participants receive an attendance record for your AI literacy file.
- Teaching time
- 6 hours on 1 day, in 2 blocks of 3 hours with a lunch break.
- Before the day
- Daily use of an AI assistant at work, or AI for Daily Work: Prompts, Assistants and Automations; no coding experience is needed.
- After the day
- A first draft of the organisation's AI Playbook: a scored use-case backlog, the methods, the rules, and a first tested build per participant.
Completion
You complete the course by working through each exercise on your own tasks. There is no exam and no grade. Each participant receives an attendance record for your AI literacy file (not a certificate).
Who is this course for?
Cross-functional AI leads, change managers and operations managers who are asked to move AI from pilots into daily work. Participants should use an AI assistant at work every day or have taken AI for Daily Work: Prompts, Assistants and Automations. No coding is needed.
At your office, send 1 or 2 people from each function that will own a use case, plus someone from IT and someone from privacy or risk.
Choose AI Basics at Work if participants are new to AI. Choose Leading AI Adoption: Governance, Policy and Culture if you need the policy and tool list without the build work.
This course is not part of a path. Book it on its own.
- Level
- Advanced
- Before you start
- Daily use of an AI assistant at work, or AI for Daily Work: Prompts, Assistants and Automations; no coding experience is needed.
- Group
- 1 to 12 people, online, live or at your office
This course is right for you if you
- lead AI use across several teams or functions
- have AI pilots that never reached daily work
- want one written way to select, build and check AI use cases
Unsure whether it fits? Ask us. We reply within 1 working day.
Practical information
- 1 day, 6 teaching hours, with a lunch break
- Online: live in 2 sessions of 3 hours on one day, not recorded; times agreed in the intake call
- At your office: 09:30 to 16:30; you provide a room, a screen and internet access
- In English, in the AI tool your organisation has approved
- Taught live by a practitioner with more than 15 years of commercial and technical leadership in software and data companies across Europe and Asia-Pacific, including maritime, energy and fintech.
- Attendance record for your AI literacy file
- Online, live: 1 day in 2 live sessions of 3 hours, not recorded. Book it as a private day for your team of 1 to 12 people, on your own Teams, Zoom or Meet.
- At your office: 1 day (6 teaching hours) for 1 to 12 people, travel within the Netherlands included.
Dates and booking
Choose the format, an open weekday in the calendar and the number of participants, then reserve on this page.
The calendar and booking form load with JavaScript. Without it, ask us for a date on WhatsApp, by phone on +31 6 20 51 73 96 or at advisory@ai-business.school.
Investment
| Item | Online, live | At your office |
|---|---|---|
| Price | €399 per person | €2,900 for the day, for your group |
| Including 21% Dutch VAT | €482.79 per person | €3,509 |
| The training day | Included | Included |
| A 30-minute intake call on your tools and data rules | Included | Included |
| Course materials and templates | Included | Included |
| Attendance record for your AI literacy file | Included | Included |
| Travel within the Netherlands | Not needed | Included |
| Group | 1 to 12 people | 1 to 12 people |
Prices exclude VAT. Online, live, you pay per participant. At your office, one price covers your whole group, from 1 to 12 people; for a larger group, book a second day. You reserve on this page; the payment details appear when you do.
Questions
What is an AI playbook and what goes in it?
A written document of how your organisation selects, builds, checks and runs AI use cases. In this workshop it holds the scored backlog, the briefing and verification methods, the data rules and governance, and the first tested builds.
What do participants receive at the end of the AI Playbook Workshop?
The organisation gets a first draft of its AI Playbook, and each participant leaves with a first tested build of one use case. There is no exam or grade, and the workshop leads to no formal qualification or professional designation. Participants receive an attendance record for your AI literacy file.
Do participants need coding skills to build AI automations?
No. Builds use your organisation's approved AI tools. A participant who uses vibe coding keeps the result as a prototype until someone who can review the code has checked it.
How long does the AI Playbook Workshop take?
1 day of 6 teaching hours. Online, it runs as 2 live sessions of 3 hours on the same day. After the workshop, the use case owners finish and test their builds.
Does the workshop support our AI Act AI literacy duty?
It supports your Article 4 AI literacy duty, and the playbook records who was trained on which tools. The AI Act does not require a certificate.
Can we choose the use cases ourselves?
Yes. The use cases come from your organisation and are scored in the first module against value, data, risk and how easy the output is to check.
Last reviewed 21 October 2026.
Related courses
AI for Daily Work: Prompts, Assistants and Automations
Build a tested assistant or simple automation for a recurring task, with a check step and an owner.
€399 per person online · €2,900 a day at your office, up to 15 people
Leading AI Adoption: Governance, Policy and Culture
Draft your AI use policy and approved tool list on your own data, and set who decides on new AI tools and uses.
€399 per person online · €2,900 a day at your office, up to 12 people
AI Strategy and Change for Decision-Makers
Decide where AI adds value and where it adds exposure, apply your sector rules, and leave with a board decision list.
€399 per person online · €2,900 a day at your office, up to 12 people