The core AIxH curriculum (how LLMs work, prompt engineering, ethics and governance, including the confidentiality and GDPR rules) is set out on the programme page. The ChatGPT module then follows a progression built for the workshop: start with the raw tool and turn it, level by level, into an assistant that knows your context, all the way to agents.
What the tool covers, beyond chat
ChatGPT goes well beyond text conversation: built-in web search, image generation, file reading, voice mode. Of all the tools in the programme, it is the one that covers the most ground on its own, while Claude and Gemini each go deeper on a narrower use.
Automating repetitive tasks
Template emails, meeting minutes, first drafts of reports: identify the tasks that come back every week in the team and build a reusable prompt for each of them.
Memory and custom instructions: the assistant knows you
Memory carries context from one conversation to the next; custom instructions set the expected tone and format once and for all. The workshop configures both, shows what memory actually keeps and how to control it.
Style guide and skills: fixing the form and the method
A style guide teaches ChatGPT to write the way the company writes, from a corpus of existing texts. Skills formalise reusable know-how: a working recipe the assistant applies the same way every time, from handling a difficult email to generating an FAQ.
Building a custom GPT for a recurring task
ChatGPT lets you create a configured version of the tool for one precise use (a “custom GPT”): fixed instructions, a set tone, reference documents loaded once. The workshop builds one for a need identified in the team, with no technical skills required, and shows how to assemble its knowledge base with deep research.
Projects: a workspace that keeps the context
A project brings together instructions, documents and conversation history for ongoing work. Where a GPT is a specialist you brief for one assignment, a project is a colleague who knows your context and remembers it. The workshop shows when to choose one over the other.
Agents and connectors: the AI that gets things done
The last level: the agent, able to carry out a task on its own (browsing, filling in a form, preparing a deliverable), and the connectors that link ChatGPT to your inbox or calendar. With the safeguards to put in place before delegating: which scope to give it, which human checks to keep.
The features few users know about
Study mode, which leads you to understanding through questions rather than ready-made answers, and conversation branches, for exploring a variant without losing the original thread. Two simple settings that change how you work with the tool.
Analysing files and data
Drop in an Excel spreadsheet, a PDF or a data export and get an analysis, a chart or a summary, without writing a formula or a line of code. Participants bring their own files.