One method for the whole team. Deeper tools for the teams that want them.

What you asked for

Practical AI, used responsibly, by everyone

Policy papers, reports, research summaries and communications: the work is demanding, the team is spread across countries, and most colleagues are just getting started with AI.

• An audit first. Before any session, we look at how colleagues use AI today, which tasks take the most time, and which rules apply. The programme is built on what we find.
• Method, not a single product. We teach how to brief an AI assistant, check its output and build reusable prompts. The method works in any assistant, including the tools EDF already uses.
• Responsible by design. Confidentiality, GDPR, the AI Act and accessibility are part of every session, not a slide at the end.
• No slide days. Every session is hands-on, on public or anonymised EDF documents.

The programme at a glance

The modules are chosen after the core programme, so each team decides with real experience of what it needs.

Core programme · AI Foundations

The whole team, the same method

One full day, hybrid on Teams. Exercises run in mixed pairs of in-room and online colleagues, every instruction is also written in the chat, and an EDF colleague acts as room host, so the four colleagues abroad take part as fully as everyone in Brussels.

Morning · The method

• What AI is good at, and where it fails. Invented facts and sources, outdated information, overconfidence: how to spot them.
• A prompting method that sticks. Role, context, task, format, examples, iteration.
• Checking before using. How to review an output before it goes into a policy paper, a report or a public statement.
• The review council. One prompt that makes the AI argue with itself: a skeptic checks the facts against the source, a realist reads it as an MEP or a journalist would, a builder proposes fixes, a chair gives the verdict. Nothing goes out unchallenged.
• Practice on EDF work. Summarising a research document and drafting a briefing note from it.

Afternoon · Responsible use and shared tools

• Confidentiality and GDPR in practice. What never goes into an AI tool, and how to anonymise what does.
• Accessible content with AI. Plain-language and easy-to-read versions, image descriptions, captions, and the limits to watch for.
• The EDF context file. The team's shared memory: who EDF is, tone of voice, terminology, current positions, audiences and accessibility rules, in one document every colleague gives the AI before starting. Same context for everyone, whatever the tool.
• The EDF prompt library. Built together during the session: reusable prompts for the team's recurring tasks, organised by role, each one starting from the context file.

Managers' briefing · 90 minutes, online

• For the five managers. Internal rules for AI use, quality and accountability, and how to support the team after the training.

Exercises run in ChatGPT, which colleagues already use. Everything learned transfers to any AI assistant EDF adopts.

Specialist modules · to choose later

Going further, team by team

After the core programme, each team can decide whether to go further. Modules B and C run in a dedicated Claude environment that we provide for the training, used only with public or anonymised documents, so no confidential data leaves EDF and no new tool has to be approved to get started.

Module A · Policy & Projects Lab

• For the policy and project team. On your own documents: drafting and reviewing policy papers, comparing positions, summarising research and consultations, preparing project reports.

Module B · Comms Studio, built around Claude

• From one document to every channel. A policy paper becomes a press release, a newsletter item and social posts, in EDF's tone of voice.
• Accessible by default. Every output comes with its plain-language version, image descriptions and captions.
• Reusable skills. Claude skills that keep EDF's style and accessibility standards consistent, whoever writes.
• A living shared memory. The context file becomes a memory Claude reads and updates: new stakeholders, changed positions, decisions taken. No one has to keep it current by hand.
• The full review council. Four separate agents review every key output before it goes out.

Module C · Automation Lab, built around Claude

• One recurring task stops being manual. Your team picks a job that takes hours every month: the press review, monitoring EU consultations, the monthly report.
• Designed together, built by us. We draw the automation together in the room and finish the technical build in the following two weeks.
• Guaranteed. If it doesn't run at the end of the build, we finish it.

Accessibility

How we deliver matters as much as what we teach

• Live captions in every session, in the room and online.
• Materials in advance, in screen-reader-friendly formats.
• Nothing only on screen. Everything shown is also said and shared, with a pace and breaks designed for hybrid participation.
• Further requirements, such as sign language interpretation, agreed during the audit.

The AI Act, handled

The training is also the evidence

Since February 2025, Article 4 of the AI Act asks organisations using AI to ensure a sufficient level of AI literacy among their staff, and to be able to show what they did. This programme produces that evidence as a by-product: an attendance register with names and dates, an AI literacy certificate for each participant, and a draft of internal AI use guidelines.

One programme, two results: a more confident team, and a file you can put on the table if anyone ever asks.

What you keep

• The EDF context file, the shared memory the whole team works from.
• The EDF prompt library, organised by role, with the review council prompt.
• A one-page draft of internal AI use guidelines, to adapt with your management.
• AI literacy certificates for every participant.
• For modules B and C: the skills and automations built during the module.

Before we start

A few questions for the audit

Tick what is already true for EDF. Your answers tell us where to start, and nothing here is required in advance.

0 / 7 checked

Investment

Clear numbers, one invoice

The modules are optional and can be added after the core programme. If EDF agrees to a short published case study, we can discuss founding partner conditions.

Prepared by White Rabbit Brussels for the European Disability Forum. Sessions run hybrid, at your offices or ours in Brussels, with colleagues abroad online, in English. Available from the week of 9 November 2026. This proposal is valid for 30 days. Full programme details: sprints.brusselswhiterabbit.eu