Skip to content
Free and open to everyone. The lessons keep improving thanks to your feedback.
Language : EN

Scaling down, stopping, asking for help

In testingChecked on 24/10/2026 (French version)Translated on 27/10/2026

This lesson is still being tested: it improves with learners' feedback. If something is unclear, it is the lesson that needs fixing, not you.

Follow this lesson with Claude Code

With the course companion installed, open Claude Code in your working folder and type:

/formation-ia:suivre m06-l05 en

It reads the up-to-date lesson on this site and guides you step by step, without changing anything without your consent. Install the companion. Claude Code is not included in the free plan: an eligible Claude subscription, or a Console account billed by usage for the terminal.

By the end of this lesson, you will be able to

  • recognize the signs that a project needs a change of plan;
  • scale down a project without sinking it;
  • stop cleanly, without guilt, when it is the right decision;
  • know who to ask for help: yourself, the AI, a person, a professional.

Why it matters to you

Many projects end neither finished nor abandoned: they drag on. Each session costs a little more energy, and produces nothing. Knowing how to scale down, stop or ask for help at the right time is what makes the difference between a project that bogs down and a project that gets finished, even smaller than planned.

Before you start

You needYour assistant, free account, or Claude Code; your project sheet
Cost€0 with a free account; Claude Code, if you use it, requires a subscription
Duration30 minutes
ReminderFor health, law and money: the AI prepares, a professional decides (module 1, lesson 2)

Understand it in 3 minutes

The signs that you need to change the plan:

  • two sessions in a row without a result you can show;
  • a step that keeps swelling (“one more session and it’ll be good”, three times);
  • the AI going around in circles: it repeats a mistake you already corrected, or always suggests the same thing;
  • a missing piece of information that blocks everything;
  • the desire fading away;
  • a subject that is beyond you: a question of health, law, money or safety.

Four possible decisions:

DecisionWhenWhat you write in the sheet
Scale downThe project is too big for the time leftWhat you remove, and why; the new minimum version
Break down differentlyA step keeps swellingThe step cut in two, each part with its check
Ask for helpInformation is missing, or the subject is beyond youWho to ask, what question, by when
StopThe project no longer serves a purpose, or costs more than it bringsThe reason, what is reusable, and putting it away in the archive

Who does what:

WhoFor
YouDecisions, anything that commits you (sending, paying, signing), anything that requires knowing the people
The AIDrafts, breakdowns, comparisons, preparing questions
Someone around youAn opinion on what is made for them, information you can’t get on your own
A professionalHealth, law, money, safety: where a mistake is costly

An AI going around in circles won’t get out of it alone. It can sound sure of itself and still be wrong. Change something: start a fresh conversation with your sheet, give it the missing information, or ask someone.

Do it together

1Step 1: Take stock

Here is my project sheet: [paste it]. I'm not moving forward. Help me take stock: which step is blocking, why in your opinion, and what are my options (scale down, break down differently, ask for help, stop)? Don't decide for me.

2Step 2: Scale down, if it’s the right option

Suggest a scaled-down version of the project, that can be finished in [number] sessions, keeping what is most useful. Tell me what we lose by scaling down.

Choose, then write the decision in the sheet: “[date]: project scaled down to [new version], because [reason].”

3Step 3: Prepare a request for help

I need to ask [a person, or a professional] for help on [the blocking point]. Help me prepare 3 precise questions, and the list of what I need to have on hand. Don't answer the questions yourself.

For a health, legal or money subject, the answer will come from the professional or from an official source, not from the AI. To prepare your questions, describe the situation in general terms, without your health information or your numbers (your “never” list).

4Step 4: Stop cleanly, if it’s the right option

I'm stopping this project. Help me write 3 lines for the sheet: the reason for stopping, what was done, and what could be useful again. Don't change anything without my yes.

Your turn

On your project:

  1. take stock with step 1, even if everything is going well (replace “I’m not moving forward.” with “I want to take stock, even if everything is going well.”, and “which step is blocking” with “is there a step that risks getting blocked”): it’s good practice;
  2. make a decision (carry on, scale down, break down differently, ask for help, stop);
  3. write it in your sheet, with its reason;
  4. if you ask for help, prepare your 3 questions before writing to the person.

Check your work

It worked if:

  • your sheet contains a decision from this review, with its reason;
  • if you scaled down: the new minimum version fits in your sessions;
  • if you ask for help: your questions are precise, and you know who to ask;
  • if you stop: the reason and what is reusable are written down, and the project is put away.

Common mistakes:

  • carrying on “one more session”, indefinitely;
  • seeing scaling down as a failure;
  • letting the AI settle a question of health, law or money;
  • giving up without writing anything: the project comes back to haunt your list.

Keep it and pick it up later

  • The decisions to scale down or stop are the most useful to reread at the review (lesson 6).
  • Keep the “take stock” request in your library (module 2).

Going further

optional
  • OpenAI acknowledges it for ChatGPT: the model can seem very sure of itself, even when it is wrong. When an AI goes around in circles, trying again identically rarely helps: fix what is missing in the request. This is also the advice of Anthropic’s course “AI Capabilities and Limitations”: a targeted correction rather than a simple new attempt.

New words

3 words explained
WordWhat it means
Take stockStop to look at what is blocking, and choose a decision
Scale downKeep a smaller version of the project, that can be finished in the time left
Go around in circlesWhen the AI repeats a mistake or the same suggestion, despite your corrections

Check your understanding

3 questions, no grade. An explanation appears after each answer.

1Two sessions in a row, your project hasn't moved forward. What do you do?
2Your project involves a rental contract, and the AI gives you a legal interpretation. What do you do?
3You stop a project before the end. What do you keep?

What's next?

Lesson 6 · Reviewing your project: measuring what the project gave you, and choosing what comes next.

Module 6:

✏️ Do the exercises

My space: your progress, your exercises and your last lesson.

Sources and checks

3 sources

Date, version, sources Written on 24/10/2026. This lesson works with a free assistant (free plans checked on 24/10/2026 on claude.com/pricing and chatgpt.com/pricing) or with Claude Code. The sentence about the AI’s confidence comes from OpenAI’s help page (“Does ChatGPT tell the truth?”), read on 24/10/2026. The role of professionals for health, law and money takes up module 1 (lesson 2).

Sources checked: