Your team has AI access. Now what?

Your team uses AI. Who checks the sources? Who judges the results? I show teams in companies, educational institutions and newsrooms how to build AI into their work without blindly trusting the output. My craft comes from journalism: checking sources.

Yvonne Pöppelbaum

Journalism-grounded teaching for AI-assisted research and AI literacy

AI is not an oracle. AI is a tool.

If you want to use it, you have to steer it with care. That's why my work starts before the tools: with the tasks your team actually has. What AI can do belongs in the training. So does what it can't. Judgement stays with people.

Four ways I work with teams

AI literacy workshops

The EU AI Act requires companies and organisations that use AI to promote AI literacy among their staff (Art. 4). I build the workshops around your team's actual tasks: what AI can do, where it fails, how to check its results and what data protection and regulation require. In German or English, on site or online.

Research and verification training

Half a day to several days: newsrooms, research teams and educational institutions work on their own research tasks and learn how to check sources, spot hallucinations and document evidence properly.

Designing learning programmes

Blended learning programmes running over several months: webinars, assignments between sessions, live sessions. I build the arc from needs analysis to evaluation.

Facilitation and moderation

Workshops, webinars, community formats. I shape the space so that groups with different levels of prior knowledge get down to work.

My reference points come from newsrooms: pressure to get the facts right, deadline pressure, healthy scepticism about AI. That's exactly where my teaching starts.

“Before, our team only used AI as a better search engine. Now we know how to use it in a structured way: check first, then let it work, then review. You can clearly tell there is a solid method and journalistic accuracy behind it. Since then, we have been working together noticeably more systematically as a team. We will definitely come back to Yvonne as soon as we're ready for the next step. Highly recommended.”
University of Tübingen, team
Institute of Media Studies
Translated from German

Order makes the difference

Between the steps, a person makes the call.

Let AI produce straight away and you get plausible text full of gaps. That's why I work in four steps: analyse first, then check and decide, then produce, then test against criteria. This sequence is part of every training I run. And of this website: it was made exactly this way.

Four things to try

I build small tools, each carrying one learning idea. Have a go.

Start screen of Question Deck: language choice Deutsch/English, below it topic filters such as Verbindung, Reflexion & Lernen, KI & Arbeit and Corporate Learning, the start button at the bottom

Question Deck

Question cards to get conversations going, built for the Corporate Learning Community Hamburg. The idea behind it: groups work better once they have talked to each other. Connection comes before content.

→ Try it

Start screen of PromptCoach: an input field for a prompt, below it six warm-up practice prompts, at the top the six steps prompt, self-diagnosis, diagnosis, revision, blind judgement, conclusion

PromptCoach

A practice workshop for prompts. I enter a prompt, the tool shows the gaps against a fixed framework and I rework them. At the end I compare both results blind and decide for myself which one works better.

→ Try it In German. Calls Claude live. You'll need to sign in briefly to use it.

Results view of the AI use case assessment: on the left seven completed steps from value contribution to result, on the right the risk category ‘minimal risk’ with notes on system type, classification and deployer role

AI use case assessment

I describe an AI project in seven steps. The tool places it in the risk categories of the EU AI Act and turns it into a briefing for stakeholders. The idea behind it: classify first, then build. Judgement stays with people.

→ Try it In German.

Start screen of the Security.Awareness.Check: a reflection game with two modules, classic security awareness and AI risks at work, five scenarios each, plus a bonus round, the ‘Let’s go’ button at the bottom

Security.Awareness.Check

A reflection game: 13 situations from everyday work, from phishing to AI risks. No score, no leaderboard. After each answer you get some context and a question to carry into your own working day. The guiding idea: AI works. People judge.

→ Try it In German.

Portrait of Yvonne Pöppelbaum: short brown hair, dark glasses, petrol-coloured polo neck jumper
Moin aus Hamburg

From journalism to learning design, down the AI rabbit hole and back

PROFILEMOIN

I help people without a tech background use AI in their everyday work. Trained journalist and experienced developer of learning programmes, workshops, AI maker days and blended learning formats. Co-lead of the Corporate Learning Community's Hamburg region. Completed in 2026: full-time training as an Agile AI Manager (DEKRA/WBS) & AI DONE RIGHT (Barbara Lampl). I like automation, I'm interested in security awareness facilitation and I keep looking for smart solutions. Since 2026, as YP Learning Design for AI Adoption, I have been supporting teams with AI adoption and agile transformation. Topics: AI for research, security awareness for journalists, advanced prompting, Claude Code and GitHub.

WHAT I'M GOOD AT

  • AI workshop design & delivery (hands-on)
  • User enablement structures & learning paths
  • Blended learning programmes (5+ months)
  • GDPR awareness & regulation
  • Advanced prompting & use case discovery
  • Process analysis & project management
  • AI governance & EU AI Act (DEKRA)
  • Learning tools, dashboards and automation with Claude Code

LIVE AND IN COLOUR

2 Dec 2026

Security awareness facilitation with a simulation game: four special units, one hijacked account, one countdown. Plus a look behind the scenes at how the material was made with Claude. In German.

5 Nov 2026

Talk “Take Credit for Your AI-Assisted Work”.

Sketchnote of the talk “How to take credit for your AI-assisted work”: on the left the problem (an experiment with identical CVs, competence penalty, usage gap), in the middle the double bind, on the right five sentences to say instead of “AI helped me”, at the bottom the closing line “Sovereignty is not a feeling you wait for. It’s a speech act.”
Sketchnote: Claudia Berg
14/15 Oct 2026
Code Week Hamburg, Tüftelhafen

Workshops for students and teachers: „Sicher im Alltag mit KI – Kompaktkurs für Lehrkräfte“ (“Safe everyday use of AI: a compact course for teachers”), „Der Anruf: Cyberangriff als Planspiel“ (“The call: a cyberattack simulation game”) and „Phishing-Detektive“ (“Phishing detectives”), also a simulation game. In German.

Code Week Germany 2026 participation badge

7 Oct 2026
AI Unconference Berghs Stockholm in Berlin
1 Oct 2026
Female AI Barcamp, Hamburg Media School

Session „Sag nicht ‚Die KI hat das gemacht‘“ (“Don't say ‘the AI did it’”) with Susanne Molter and Nicola Peschke: how do we phrase things so that our expertise shows? In German.

Sketchnote of the session “Don't say ‘the AI did it’” at the first Female AI Barcamp at Hamburg Media School: on the left, sentences for naming your own use of AI, such as “I define topics and structure, the AI suggests content”; in the middle, a woman and a robot at a laptop with the question “Who decides over whom?”; on the right, clouds about the future in 2036 with the questions “Desirable?”, “Which assumptions about women, AI and work lie behind our images of the future?” and “What can we do today?”
Sketchnote: Ania Groß
12/13 Sep 2026
AI.Women Hackathon, Kühne Logistics University Hamburg

Mentor at the largest AI hackathon for women in Germany (according to AI.Women), with 300 participants (GitHub, Claude Code, prompting, product development)

Sep 2026
University of Tübingen

AI enablement focused on Claude Projects for the team of the Chair of Media Studies.

Organisations I have worked for as a freelancer in recent years: Weltreporter (co-management), Goethe-Institut, FES Medienakademie, University of Tübingen, Lie Detectors, Spotlight Verlag (now ZEIT Sprachen), managerSeminare, Unicum Verlag, telc, Münchner Stadtbibliothek (Munich City Library) and Rautenstrauch-Joest-Museum Cologne. Most recently I was employed by tactile.news in Lüneburg (workshops and consulting) and by Freischreiber (co-management).

CERTIFICATES & FURTHER TRAINING

YearCertificateIssued by
2026AI DONE RIGHTBarbara Lampl
2026DEKRA Agile KI-Manager*in (Agile AI Manager; incl. AI governance, EU AI Act, ISO 42001, AI risk, PSM I and PAL I)WBS Training / DEKRA / Scrum.org
2026Advanced Prompt Engineering for JournalistsKnight Center for Journalism in the Americas (Joe Amditis)
2025AI & Learning Design Bootcamp – FRAME™ methodologyMaven / Dr Philippa Hardman
2024/2025Designing & Facilitating Learning Spaces I + IIKAOSPILOT
2024Generative AI for Media Professionals MasterclassMaven (Jeremy Caplan & Nikita Roy)
2019Train-the-TrainerReporter-Akademie, Berlin

Let's talk about your team

Tell me what your team is working on right now. I'll reply with a concrete proposal.

info@poeppelbaum.eu

→ LinkedIn

↑