Heartist Journey 2026
Workshop 2 recap
Presented by
Przemek BlicharskiAI & Data Literacy TrainerHR teams and GMs now have a new job: helping their teams use AI safely. Write clear prompts (start with the role) and improve the answer step by step. The most useful tool of the session was NotebookLM. It answers only from your own documents and shows where each answer comes from.
The main messageAI is coming whether we like it or not, so prepare. Give people safe tools and good habits. Humans keep the decisions and the final check.
7 short parts, in order. Swipe to see them all, and tap one to go there:
Part 1 of 7
Przemek, a psychologist who has worked with AI since 2014, used a fictional hotel and fake data, so nobody needed real company information. The hotel’s three problems: writing from zero, the same questions again and again, and documents nobody reads.
Part 2 of 7
HR and GMs are now “AI officers”, whether they like it or not. The EU AI Act asks for AI training, and people already use AI. Without safe tools, they put company data into private apps.
AI has real costs, like energy use, but it is coming anyway. Better to be ready.
Part 3 of 7
AI predicts the next word, like your phone’s autocomplete. It does not fill the gaps like a colleague does, so be specific. People keep the decisions, the difficult conversations and the final check.
Part 4 of 7
A good prompt has six parts: role, context, task, format, tone and limits. The role matters most.
Tip: save one chat that writes good prompts for you. And always ask again, because the first answer is rarely the best.
Part 5 of 7
NotebookLM answers only from the documents you upload, with clickable sources, so it does not invent answers. In the demo, it found that a hotel handbook broke Polish overtime law.
It can also turn documents into slides, quizzes, infographics or a short audio podcast.
Part 6 of 7
Create an FAQ notebook from your policies and share it with your team. Use it to prepare audits or to check handbooks against local law. A simple automation can even send you a news email every morning.
Part 7 of 7
Practise with fake data. Never put guest or employee data into public AI tools, and use Accor’s approved tools.
AI Tech Lab
Role, context, task, format, tone, limits. No time? Give the role.
A saved chat that turns a short request into a full prompt.
Answers only from your documents, with clickable sources. Up to 50 sources. Can make slides, quizzes, infographics and audio.
Read the answer, then ask for changes: shorter, simpler, a different style.
A news summary sent to your email every morning (personal Google account).
“Whether you like it or not, you also became an AI officer.”
“If you don’t have time for anything else, just give the AI the role.”
“We should leave as little to chance as possible when talking to AI.”
“Rarely is the first output from AI the one that is really good.”
QHow much do these tools cost?
AMost paid plans cost about $20 a month. The main NotebookLM features are free.
QCan AI send me a news summary every morning?
AYes. A simple Gemini automation in Google can send it to your email, for example HR news.
QIs it safe to upload documents to the cloud?
ABig cloud services are usually safer than a laptop. Phishing is the bigger risk. But never upload personal data, and follow Accor’s rules.
QCan I share a notebook with my team?
AYes. You can share it with chosen people or with everyone who has the link.
We wrote this recap from the session recording. Quotes are slightly edited to make them easier to read.
