AI at Work 101
People who have been handed an AI assistant at work, or are about to be, and want to know what it is before they trust it
Prerequisites: None. A laptop and access to any one of the four assistants, on any plan.
Last verified 2026-09-17
Learning objectives
By the end, participants can:
- Describe a large language model as next-word prediction at scale, and name two things that follow from it
- Recognise hallucination and sycophancy in an answer, and know the one habit that defuses each
- Explain tokens and context windows well enough to know why long chats degrade
- Tell the four surfaces apart (chat, desktop app, inside the office suite, agents) and pick the right one for a task
- Complete and iterate a first useful task on their own material, with any of the four assistants
What's in this module
7 lessons · 5 exercises · ~8 min reading
Starts withWhat a large language model is~1 minExecutive summary
You will learn what a large language model actually is, what that makes it good and bad at, and how the four assistants most organisations use (Claude by Anthropic, Gemini by Google, ChatGPT by OpenAI, and Microsoft 365 Copilot by Microsoft) package the same idea in different surfaces. End state is one real task from your own work completed with whichever assistant you have, iterated to something you would actually use. After this module you can explain hallucination and sycophancy to a colleague, tell a chat surface from an agent surface, write a first useful brief, and know when to start a new conversation instead of arguing with an old one.
Sources
- Anthropic, Claude pricing, the plan lineup and the model-training default per tier: https://claude.com/pricing
- Google, Gemini Apps Privacy Hub, what a personal account's chats are used for and the Keep Activity setting: https://support.google.com/gemini/answer/13594961
- OpenAI, ChatGPT pricing, the plan lineup, the surfaces per plan and the training opt-out: https://chatgpt.com/pricing
- Microsoft Learn, Data, privacy and security for Microsoft Copilot, the vendor's own statement that generated content is not guaranteed factual and must be reviewed: https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy
- Google Workspace, AI solutions overview, where Gemini appears across the suite: https://workspace.google.com/solutions/ai/
Prefer the live room?
This module also runs inside our in-person bootcamps and workshops in Copenhagen.
More in Track A
Claude 101
What Claude is, how it differs from ChatGPT, and how to use its three surfaces (browser, desktop, Cowork) for real work.
How to Prompt
How current frontier models follow instructions, and the five fixes that turn vague prompts into precise ones.
Choosing Your AI Assistant
How to choose between {{fact:general_assistants}} for an organisation, starting from what you already pay for rather than from benchmark tables.