Curriculum
Nine modules. The four marked core run in every session; the rest are chosen to fit the room and the time.
Each outcome is written as something you could watch a person do. If we could not phrase it that way, the module would not be here.
M1
What these things actually are
45 mincore- Covers
- A working mental model of a language model: it predicts plausible continuations, it has no database of facts, and it does not know when it is wrong. Where the training cut-off sits and why that matters. The difference between a model, a chat product and an agent.
- Afterwards they can
- Can explain to a colleague why the model invented a citation, rather than concluding it is broken or that AI cannot be trusted at all.
M2
Getting a usable answer
90 mincore- Covers
- Context, role, examples, constraints, format. Why "write me a report" fails and what to give it instead. Iterating rather than re-rolling. When to start a new conversation. Working from a template you keep.
- Afterwards they can
- Takes a task they do weekly and produces a first draft they would actually edit, not discard. Has written two reusable prompts for their own role.
M3
Checking the output
60 mincore- Covers
- Which errors these systems make: invented specifics, confident arithmetic mistakes, silently dropped requirements, and plausible-sounding summaries that omit the important clause. Building a per-task check that is proportionate to what a mistake costs.
- Afterwards they can
- Can state, for a given task, exactly what must be verified before the output leaves their hands.
M4
What must never go in
45 mincore- Covers
- Client data, health information, employee records, unreleased financials, anything under an NDA. The difference between a consumer account and a business tier on retention and training. Australian Privacy Act obligations in practical terms. What to do if someone has already pasted something they should not have.
- Afterwards they can
- Can look at a document and say whether it may be pasted into which tool, without checking with anyone.
M5
Documents, long and short
60 min- Covers
- Summarising accurately, extracting structured information from unstructured text, comparing two versions, drafting from a template. Where long documents get lossy and how to notice.
- Afterwards they can
- Produces a summary of a real document from their work and can identify what the summary left out.
M6
Spreadsheets and numbers
60 min- Covers
- Formula generation and explanation, turning data into narrative, cleaning messy exports. Critically: why you never trust arithmetic done in prose, and how to make the tool show its working.
- Afterwards they can
- Generates a working formula for a real problem and has a rule for when to recompute rather than trust.
M7
Writing that sounds like you
45 min- Covers
- Voice and tone, editing rather than accepting, when a first draft is genuinely faster and when writing it yourself is. Avoiding the flat register that makes AI-written text obvious.
- Afterwards they can
- Can spot unedited AI output in a piece of writing and knows the three or four tells to remove from their own.
M8
Spotting the use cases worth having
60 min- Covers
- A simple screen: how often does this task happen, how long does it take, how bad is a mistake, and is the input already digital. Applied to the room's own suggestions.
- Afterwards they can
- Team leaves with a ranked shortlist of their own tasks and a clear "not worth it" pile, which is usually the more useful half.
M9
Copilot in the apps you already pay for
90 min- Covers
- Microsoft 365 Copilot in Word, Excel, Outlook and Teams specifically. What it can see and what it cannot. Why results differ so much between users, and what that says about how your files are organised.
- Afterwards they can
- Uses Copilot on their own inbox or documents and understands why a colleague gets different results.
What we deliberately leave out
- The history of AI. Interesting, changes nothing about Monday.
- How the models work internally. Attention heads and embeddings do not help a bookkeeper.
- Tool tourism. Twenty products demoed in an hour is entertainment. We teach two or three properly.
- Predictions about jobs. Nobody knows, and the speculation makes rooms anxious rather than capable.
Choosing modules
A half day fits the four core modules. A full day fits core plus three or four more. Which extras depends entirely on who is in the room, which is what the scoping conversation is for.
Formats and group sizes