Using AI at work: 5 simple rules for your teams
Client data, context, checking, decisions, disclosure: the five rules of an AI use policy, in a 1 min 44 s video, then in detail with the EU rules behind them.
4 min read
Your teams already use AI assistants, sometimes on personal accounts. A five-rule policy explained in under two minutes is a concrete first step. Here are the rules, why they hold, and what they leave out. The video is in French; its narration is translated below.
Why now
In the EU, Article 4 of the AI Act has applied since 2 February 2025: organisations that use AI systems must take measures so their staff have a sufficient level of AI literacy, given their knowledge, training and context of use. Supervision and enforcement rules apply from 3 August 2026.
What the European Commission's Q&A spells out:
- an employee drafting with a ChatGPT-type assistant is covered, and should know the specific risks, such as made-up answers;
- no specific training or certificate is mandatory; internal records of what you did are enough;
- the effort depends on the systems used and their risks.
Outside the EU, the same five rules are simply good practice.
The 5 rules
1. Never put client data in a personal account
On consumer plans, your chats may be used to improve the models if you allow it in your settings. A client contract pasted into a personal account has left the company.
What to do: use the workspace your company has approved (a business plan with checked terms), and only share what the task needs.
2. Give context
| Too vague | With context |
|---|---|
| "Summarise this." | "Summarise this report in five points for our CFO." |
A good prompt answers three questions:
- Who for? The reader, and what they already know.
- What for? The decision they need to make.
- In what form? A page, a table, an email.
3. Check everything
An AI assistant can produce wrong or misleading answers, even when they sound sure. Anthropic says so in its own help pages: Claude should not be relied on as a single source of truth. Think of it as a brilliant intern: fast, but to be checked. Figures, names, quotes: always go back to the source.
4. AI suggests, you decide
The answer is a draft. You sign what goes to the client, and you answer for it.
5. Say when AI helped
Clients should know how their deliverable was prepared. A plain note is enough: "first draft written with an AI assistant, reviewed and approved by the team".
What the video does not cover
The video is about daily use. It does not replace:
- choosing and configuring the tool: France's cybersecurity agency ANSSI published security recommendations for generative AI systems in April 2024;
- a data protection review of the processing involved;
- job-specific training: drafting an audit with AI is not learned like drafting an email.
How this video was made with EducPilot
- Sources uploaded: the firm's five rules (a fictional document) and two public Claude help pages, on data use and on possible errors.
- The plan: a hook ("Monday, 9 am"), the five rules announced, one scene per rule, a recap, a closing question.
- Devices: a weak and a strong prompt side by side, an analogy (the brilliant intern), the three questions of a good prompt as a list.
- Sources in the film: each claim about data and errors links to the help page that supports it.
- Review: two rounds. A note applied at 0:14: "Name the five rules, and draw each circle on its word."
Claude is used here as an example tool. EducPilot has no partnership with Anthropic.
To do the same with your own policy, see How to turn a PDF into a training video.
The narration, translated
Monday, 9 am. A consultant pastes a client contract into his personal Claude account. He saves an hour. But that data has just left the firm.
At Helio, using Claude well comes down to five rules: data, context, checking, decisions, and transparency.
One: never put client data in a personal account. Your chats there may be used to improve the models, if you allow it. Use the Helio workspace, approved by the firm.
Two: give context. "Write a summary" is too vague. "Summarise this report in five points for our CFO": that is context. A good prompt answers three questions. Who for? The reader, and what they already know. What for? The decision they need to make. In what form? A page, a table, an email.
Three: check everything it produces. Claude can be wrong, even when its answer sounds sure. Think of it as a brilliant intern: fast, but to be checked. Figures, names, quotes: always go back to the source.
Four: Claude suggests, you decide. And you sign what goes to the client.
Five: say when AI helped you. Your client should know how their deliverable was prepared.
Remember: the Helio workspace, context, check everything, sign, and say so. Before you send, one question: have I checked everything?
Sources
- European Commission, AI Literacy: questions and answers (Article 4 of the AI Act).
- CNIL, Entry into application of the European AI Regulation: the first questions and answers from the CNIL.
- ANSSI, Security recommendations for a generative AI system, 29 April 2024 (PDF).
- Anthropic, How do you use personal data in model training?
- Anthropic, Claude is providing incorrect or misleading responses. What’s going on?
Full transcript
Original narration, in French.
Lundi, 9 h. Un consultant colle un contrat client dans son compte Claude personnel. Il gagne une heure. Mais ces données viennent de quitter le cabinet.
Chez Helio, bien utiliser Claude tient en cinq règles. Les données, le contexte, la vérification, la décision, et la transparence.
Un : jamais de données clients sur un compte personnel. Vos échanges peuvent y servir à améliorer les modèles, si vous l’acceptez. Passez par l’espace Helio, validé par le cabinet.
Deux : donnez du contexte. « Fais une synthèse », c’est trop vague. « Résume ce rapport en cinq points, pour notre directeur financier » : voilà du contexte.
Un bon prompt répond à trois questions. Pour qui ? Le lecteur, et ce qu’il sait déjà. Pour quoi ? La décision qu’il doit prendre. Sous quelle forme ? Une page, un tableau, un e-mail.
Trois : vérifiez tout ce qu’il produit. Claude peut se tromper, même quand sa réponse semble sûre.
Voyez-le comme un stagiaire brillant : rapide, mais à relire.
Chiffres, noms, citations : remontez toujours à la source.
Quatre : Claude propose, vous décidez. Et c’est vous qui signez ce qui part chez le client.
Cinq : dites-le quand l’IA vous a aidé. Votre client doit savoir comment son livrable a été préparé.
Retenez : l’espace Helio, du contexte, tout vérifier, signer, et le dire.
Avant d’envoyer, une seule question : ai-je tout vérifié ?
Try it with your own documents
Upload a PDF or a procedure, approve the plan, correct the video by annotating it.
Read next
- How to turn a PDF into a training video with AI: method, examples and limits
From a 40-page procedure to a 2-minute video people actually watch: a six-step method, what AI does well and badly, and real examples generated from official documents.
- Whiteboard explainer videos for corporate training: when they work, how to make one
Whiteboard video helps people follow reasoning, not copy a physical task. When to use it, when to film instead, what research says, and how to make one.
- How to write a training video script: a situation, one idea per sentence, sources
How to script a one to three minute training video: open on a situation, one idea per sentence, write for the ear, source every fact. With a real annotated script.