The Five-Minute Setup That Makes AI Actually Useful for Teachers
If AI outputs feel 'fine but not quite you', it's usually missing context. A simple Project (ChatGPT) or Gem (Gemini) stores your exam board, class, tone, and preferences so prompts become one sentence — and resources come out classroom-ready.
Quick Summary
- •Most 'bad AI output' is actually 'missing context' — exam board, class, tone, and formatting.
- •A Project/Gem lets you store your teaching world once so you stop repeating yourself.
- •With context saved, prompts can be one sentence and still produce spec-aligned resources.
- •Updating a Project takes seconds and pays back all year in reduced rework.
Summary: If your first experience with AI produced something that was fine but not quite right — wrong tone, slightly American examples, shaky exam alignment — the problem usually isn't the model. It's the missing context. A five-minute Project (ChatGPT) or Gem (Gemini) fixes that by storing your teaching world once, so everything you ask for comes out far closer to classroom-ready.
Why AI output often feels "fine" but not you
Most of us have had the same first experience with AI.
You ask it for a worksheet or a model answer, and it produces something… fine. Not awful, but not quite you.
The tone is slightly off. The examples feel a bit American. The difficulty level drifts. The exam board alignment is shaky.
It's easy to conclude the model just "isn't good enough".
In my experience as both a classroom teacher and while building Teach Edge, the issue is usually not the model. It's the context.
Once the AI knows who you are, who your students are, what you're teaching, and how you like things done, the whole experience changes. Suddenly the output feels like it came from a colleague in your department.
What "context" actually means in a school setting
When teachers talk about context, we don't just mean the topic title. We mean the things that shape every decision we make:
- the exam board and specification you teach
- the year group, age, and ability
- your tone when you talk to pupils
- the style of resources your school prefers (Word not Google Docs, print-friendly, no icons)
- your class list (for grouping and targeted tasks)
- the common misconceptions you see every year
- the real examples you return to again and again
A general AI model doesn't know any of that unless you tell it — and repeating it every time is what makes AI feel fiddly.
This is exactly what Projects in ChatGPT (or Gems in Gemini) quietly solve.
They let you embed your teaching world into the AI once, so you never have to re-explain the basics.
How a Project transforms day-to-day planning
Once the context is stored, here's what changes.
1) Worksheets match the spec first time
No more correcting AQA concepts in an Edexcel lesson. No more American case studies creeping in.
You get:
- British examples by default
- the right command words
- the right assessment focus (AO alignment)
2) Group tasks slot straight into your class
If your Project stores your student names (or pseudonyms), you can ask:
"Put them into four groups and assign each member a sub-topic."
It simply happens. No copying names across. No forgetting who works well together.
3) Your tone stays consistent
You don't have to keep reminding the AI:
"Warm, British, teacher-to-teacher. No gimmicks."
It's already there. Everything sounds like you wrote it.
4) You can iterate like you would with a colleague
Once the baseline is right, your follow-ups can be quick:
- "Make it a bit easier for my lower-achieving students."
- "Add a real-world example from tech."
- "Turn this into a 20-minute revision task."
The AI adjusts without losing the thread.
5) Most prompts become one sentence
Once the Project is set up, prompts like these suddenly work:
- "Turn this idea into a worksheet for tomorrow."
- "Give me a model answer for this past paper question."
- "Create a 10-minute starter on price elasticity."
And it comes out in your voice, for your exam board, for your class.
How to set up a good AI Project (in five minutes)
You don't need to overthink it. Add the things you find yourself repeating anyway.
Step 1: Add the teaching essentials
- exam board
- year group
- current topic
- subject conventions (AO structure, command words, typical question types)
Example:
"Year 13 Edexcel Economics A Level. Currently teaching Theme 3: microeconomics. Use assessment objectives AO1 (knowledge), AO2 (application), AO3 (analysis), AO4 (evaluation)."
Step 2: Add your classroom preferences
- voice and tone
- formatting rules
- resource types you prefer
Example:
"Warm, clear, teacher-to-teacher tone. Always create print-friendly resources with space for answers. No icons or clipart."
Step 3: Add your class list
This is what unlocks quick groupings, targeted practice, and personalised revision.
Example:
"Class list: Aisha, Ben, Chloe, Dev… (initials are fine)."
You can add short notes if you want, but even just names is useful.
Step 4: Add recurring examples and misconceptions
This is the "department handbook" bit — what you'd tell a trainee teacher.
Example:
"Common misconception: confusing supply shifts with movements along the curve. Useful examples: Uber surge pricing, petrol prices after OPEC decisions, post-pandemic housing demand."
Keep it updated
When your class changes topic, update the Project. When you spot a new misconception, add it.
It takes seconds. It saves hours.
Examples: before and after using a Project
Before (no context)
Prompt: "Make me a labour-market worksheet."
Result: generic worksheet, wrong exam board, American examples, and no space for answers.
After (with a Project)
Prompt: "Make me a labour-market worksheet."
Result: Year 13, Edexcel-aligned, UK examples (NHS pay disputes, zero-hours contracts), correct AO focus, warm teacher voice, print-friendly layout with answer space.
It feels like it came from the department.
The bigger point: AI should feel like a colleague, not a wildcard
When AI knows your context — your subject, your class, your voice — it stops feeling like a novelty tool and starts feeling like part of your workflow.
It helps you get ahead rather than fire-fighting.
It reduces cognitive load instead of adding to it.
And most importantly, it frees you up for the bits of teaching only a human can do: the moments in the room, the questions that catch you off guard, the student who finally gets it.
If you haven't tried building a Project or Gem yet, start small: exam board, class, tone. Add the rest over time.
It's a five-minute setup that pays back all year.
Gary Roebuck is Head of Economics at Holy Cross School, New Malden and the creator of TeachEdge.
Related Posts
An Introduction to AI Prompting for UK Teachers
AI prompting (sometimes called prompt engineering) is simply the skill of giving clear instructions so you get useful, teacher-ready output. This guide shows a practical, UK-teacher-friendly way to structure prompts that actually work.
The UK Teacher's Guide to AI in Education
A practical handbook for UK teachers who want to use AI to reduce workload and enhance teaching — without losing professional control. Includes GDPR-safe habits, classroom prompts, and a simple first-week plan.
Beyond Marking: How UK Teachers Are Getting Creative with AI
AI tools like ChatGPT can help with marking and feedback — but many UK teachers are using them for far more: lesson planning, classroom engagement, differentiation, communication, and personalised support.
Ready to transform your marking workflow?