In most classrooms, students see polished answers far more often than they see messy, in-progress thinking. Yet success in modern curricula and future workplaces depends on being able to break down complex problems, test ideas, revise plans and justify decisions.
AI tools like Claude can now produce long, structured chains of reasoning. Used well, this is not about giving students instant solutions. It is about slowing thinking down, making invisible processes visible, and giving learners a safe space to practise multi-step reasoning with support.
Extended thinking is especially valuable when you want students to:
If you are interested in how this fits into broader skill development, you might also like future-proofing students’ skills AI can’t replace.
In classroom terms, Claude’s extended thinking is similar to hearing a very articulate student “think aloud” while solving a problem on the board. Instead of just giving the final answer, Claude can:
For example, in a history lesson you might ask:
“Explain why the Treaty of Versailles caused political tension in Germany. Show your reasoning step by step, as if you are a student planning a paragraph, not just giving the final answer.”
Claude might then outline key factors, connect them to evidence, and sketch a paragraph structure. This is chain-of-thought in a form students recognise: like a detailed plan on a mini whiteboard, not a finished essay.
In a science lesson, you might see Claude:
The power lies in how transparent this process is. It gives you something concrete to critique, adapt and gradually hand over to students.
To use Claude as a worked example partner, the way you phrase prompts matters. You are not simply asking for “the answer”; you are asking Claude to model the sort of thinking you want students to copy and adapt.
You can guide this by:
For instance:
“Solve this equation and show your thinking step by step. Imagine you are explaining to a friend who is good at arithmetic but nervous about algebra. Make each step small and clear.”
Or, in literature:
“Analyse this poem’s use of imagery. First, list what you notice. Then group your observations into 3–4 big ideas. Finally, show how you would turn those ideas into a paragraph plan. Think aloud as you go.”
If you are exploring when this kind of AI support helps and when it risks harm, see when AI helps vs harms learning.
Once Claude has produced a clear chain-of-thought, you can turn it into worked examples and gradual scaffolds.
One approach is:
Full worked example
Show Claude’s complete reasoning on the board. Read it with the class, pausing to ask, “Why this step?”, “Could we do it another way?”, “What’s missing?”
Partially completed example
Remove some steps and ask students to fill the gaps. For example, keep Claude’s structure but blank out justifications or calculations.
Prompt-only scaffold
Keep only the headings from Claude’s reasoning (e.g. “Step 1: Identify known information”). Students complete everything beneath.
Independent problem solving
Students tackle similar problems with no scaffold, then compare their reasoning to Claude’s afterwards.
This “fading” process mirrors how we use worked examples in many subjects. Claude simply speeds up the creation of high-quality, step-by-step models that you can adapt on the fly.
Ready to Revolutionise Your Teaching Experience?
Discover the power of Automated Education by joining out community of educators who are reclaiming their time whilst enriching their classrooms. With our intuitive platform, you can automate administrative tasks, personalise student learning, and engage with your class like never before.
Don’t let administrative tasks overshadow your passion for teaching. Sign up today and transform your educational environment with Automated Education.
Extended thinking is most powerful when it becomes a routine rather than a one-off novelty. Here are some subject-flavoured patterns you might build.
Use Claude to generate multiple solution paths for the same problem.
Ask:
“Show two different methods to solve this problem. For each method, explain why it works and when it might be better or worse than the other.”
In class, students:
Over time, they internalise the habit of evaluating strategies, not just following the first one they see.
Use Claude to model planning, not just calculation.
Prompt:
“Plan an investigation to test how light intensity affects photosynthesis. Think aloud as you decide what to keep the same, what to change, and what to measure. Explain the reasoning behind each decision.”
Students can then critique the plan, suggest improvements, and finally design their own investigation on a different factor (e.g. temperature) using the same reasoning structure.
Use Claude as a “silent partner” for essay planning.
Ask:
“Here is a question and some notes. Show how you would turn these into a structured argument. First, sort the notes into groups. Then decide on an order. Finally, sketch topic sentences. Explain your choices as you go.”
Students can compare Claude’s plan with their own, borrowing useful structures but still choosing their own evidence and wording.
Use extended thinking to model how to build complex sentences.
Prompt:
“Turn this simple sentence into a more advanced one suitable for an exam answer. Think aloud as you add detail, connectives and more precise vocabulary.”
Students then imitate the process on different sentences, using Claude’s reasoning as a template rather than copying its final version.
For more on using AI as a co-teacher rather than a replacement, see human–AI co-pilot model teaching.
There is a real risk that detailed AI reasoning can remove too much of the struggle. If students simply follow Claude’s steps mechanically, they may look successful without developing transferable skills.
To keep productive struggle alive:
You might also explicitly discuss with students: “What parts of this thinking do you want to be able to do without help by the end of the term?”
Extended thinking is an excellent gateway into metacognition – thinking about thinking. Because Claude’s reasoning is written down, students can step back and analyse it more easily than their own internal thoughts.
You can ask:
Then invite students to annotate Claude’s chain-of-thought with their own comments, questions and alternative strategies. This helps them build a personal toolkit of approaches for complex problems.
If you are thinking about longer-term skill development, you may find openai-o1 reasoning models for educators a useful comparison of different AI reasoning tools.
Whenever you use extended AI reasoning, it is important to be clear about boundaries.
In many systems, students are not allowed to use AI tools directly for coursework or exams. Your classroom use should therefore focus on process, not product. Some practical safeguards include:
You might also explicitly discuss with older students how exam conditions differ: “Here, Claude is helping us practise planning and checking. In the exam, you will need to run that process in your own head.”
You do not need to redesign a whole unit to begin. Choose one class and one topic where students often get stuck in the middle of a problem, not at the start.
Next week, you could:
Start small, watch how students respond, and adjust the level of detail and support. Over time, Claude can become a reliable “thinking partner” that helps you make complex reasoning visible, while you remain firmly in charge of the pedagogy.
Happy reasoning!
The Automated Education Team
Classroom displays can be powerful teaching tools, but only when they are …
This ready-to-run INSET Day workshop helps staff move from general AI …
Timetabling and daily cover are high-stakes operational puzzles, but AI is …
An AI Acceptable Use Policy written once and filed away won’t keep pace …
2025–26 brings sharper, more practical expectations for how schools manage …
A back-to-school AI plan succeeds when it is small, predictable and safe …
Clearing moves fast, but students deserve decisions that are calm, …
Results season can feel like a rush of numbers, narratives and urgent …
Summer reading can slip quietly, especially for pupils who need extra …
Claude’i laiendatud mõtlemine võib toimida väsimatu „läbiproovitud näite partnerina” sinu …
Claudes utökade tänkande kan fungera som en outtröttlig “partner för genomgångna …
Claude’n laajennettu ajattelu voi toimia luokassasi väsymättömänä …