Writing workflow guide

Where does the toolkit sit in the writing process?

Use this guide to decide when to use the toolkit, which chain of tools to try, and how to move from feedback into your own revision.

Main pattern

Own attempt → diagnosis → practice → revision → record.

The toolkit is strongest after you have produced something yourself. It helps you diagnose what is happening, learn one manageable skill, revise your own work and record AI use if needed.

Before writing

Use AI cautiously at this stage. You can ask for explanation of a task or practise with a made-up example, but avoid letting AI choose your argument, structure or first wording for you. The early struggle is part of learning.

After a rough draft

This is usually the best point to use the toolkit. You have something of your own to test, and the feedback can respond to your actual thinking rather than replacing it.

NeedPossible chain
Sentence and grammar improvementWT3 → WT4 → WT1 → SW1
Paragraph clarityWT2 → WT1 → SW1
Whole-draft structureST1 or ST2 → AT2 → SW1
Argument depthAT3 → AT7 or AT9 → SW1
Research proposal developmentRP1 → RP3 → RP4 → SW2

When you receive tutor feedback

Use SW2 to turn the feedback into concrete actions. Then use the relevant writing, structure, academic-thinking or proposal tool to work on one action at a time.

Before submission

  • Check that the final wording is yours.
  • Check that your evidence, citations and claims are accurate.
  • Check your course rules on AI use.
  • Use SW3 if you need an AI-use record or declaration.

When you feel stuck

If you do not know what to do next, say “I’m stuck”. The tutor should slow down, take a step back, and help you identify a manageable next move.

A good response should not give you a long list. It should usually offer two or three possible ways forward and ask whether one of them fits.

Avoid feedback dependency

Do not keep asking the AI to review the same section until it sounds polished. A better pattern is: get one diagnosis, revise yourself, then check whether the problem has changed.

Question the feedback

Do not just accept AI feedback. Use it as something to think with, question and check. If the AI misreads your meaning or your subject context, push back and ask it to review the point again.