AI Tools for Tutors to Create Lesson Plans (Without Losing Your Sunday Evenings)

The short version: ChatGPT, Claude, MagicSchool AI, Eduaide.ai, Twee, and Brisk Teaching can each cut lesson planning time by 60 to 80 percent, but only if you feed them real detail about the specific student in front of you. Skip that step and you get a generic worksheet that any parent could spot in thirty seconds, and any child can feel in five.

Why tutors ended up needing this more than classroom teachers do

A classroom teacher plans one lesson for thirty kids. A tutor plans thirty lessons for one kid each, sometimes on the same evening, sometimes with fifteen minutes' notice because a parent has just texted to say little Freddie is struggling with fractions before Thursday's test. That's a different problem, and it's why the AI tools built for schools don't always fit tutors without a bit of bending.

I've spent the last two years watching independent tutors and small tutoring agencies adopt AI, partly through my own consulting work and partly because my daughter used a tutor through GCSEs and I sat in on more than one of those planning conversations. What I've seen convinces me most tutors are using maybe a third of what these tools can offer them, and wasting time on the bit that doesn't matter (making it look pretty) while skipping the bit that does (making it fit the child).

The tools worth your money, and what they cost in 2026

There are dozens of "AI lesson planner" products now. Most are wrappers around the same two or three underlying models. Here's what I'd tell a tutor to try, in order of how often I see them used well:

• ChatGPT (Plus, $20/month) - the most flexible option because you can describe a student's exact weaknesses in plain English and it will build around them. The free tier works fine for occasional use but hits message caps during busy exam-season weeks, which is worth knowing before you rely on it the night before a mock exam. If you want to get more out of the free version specifically, my guide to ChatGPT prompts and daily caps walks through exactly where those limits sit and how to word prompts so you don't waste your allowance on drafts you'll throw away.
• Claude (Pro, $20/month) - better at longer, structured documents, so if you're building a full six-week scheme of work rather than a single session, Claude tends to hold the structure together more consistently across a long conversation.
• MagicSchool AI (free tier, Pro at $9.99/month for individuals) - built specifically for education, with over 80 purpose-made tools including differentiation, IEP-style accommodation suggestions, and rubric generators. Worth it if you're teaching multiple subjects and want templates rather than blank-page prompting.
• Eduaide.ai (free tier limited to a small number of generations a month, paid from $9.99/month) - strong for quickly generating comprehension questions, exit tickets, and differentiated reading passages at three reading levels from one source text, which is useful for tutors working with mixed-ability siblings.
• Twee (free tier, premium around $25/month, mostly aimed at language tutors) - if you tutor English as a foreign language, Twee turns any YouTube video or article into a full lesson with vocabulary, comprehension, and discussion questions in under a minute.
• Brisk Teaching (Chrome extension, free tier, Pro around $15/month per user) - works inside Google Docs and Slides, so if a tutor already builds worksheets there it slots into the existing workflow rather than forcing a new one.

A real example: how one maths tutor cut her planning time

A tutor I worked with in Brighton, running her own GCSE maths tutoring business, used to spend roughly ninety minutes on a Sunday planning her week of eight one-to-one sessions. She'd pull past papers, cross-reference the exam board specification, and write out worked examples by hand for each student's specific gap.

We rebuilt her process around ChatGPT in one afternoon. She now keeps a running note per student (topics covered, last test score, specific error patterns, exam board), pastes that into a prompt template, and asks for a 50-minute session plan with three worked examples, one common misconception to address, and a five-question exit check. Her planning time for eight students dropped to about twenty minutes total. That's not a rounding error, that's roughly seventy minutes back every single week, and she now uses part of that time to mark the exit checks and adjust the following week's plan, which is the bit that used to fall off the end of her list entirely.

Step by step: building a week of tutoring plans in under an hour

Here's the process I'd hand any tutor starting today:

• Step 1 - Keep a one-line student profile updated after every session: last topic, score or confidence rating, one specific thing they got wrong.
• Step 2 - Open your tool of choice and paste that profile straight into the prompt, along with the exam board or curriculum stage (AQA, Edexcel, OCR, or "Year 4 UK National Curriculum" all change the output meaningfully).
• Step 3 - Ask for the plan in a fixed structure every time: warm-up, main teaching point, worked example, independent practice, exit check. Consistency here means you can scan a plan in ten seconds rather than reading it fresh each time.
• Step 4 - Ask the tool to flag "the most likely mistake this student will make" before you even start teaching. This one step is what separates a generic plan from one that addresses the child.
• Step 5 - Read it once, cut anything that sounds like it was written for a classroom of thirty rather than one child, and adjust the language level if needed.
• Step 6 - Save the good ones as templates. This is where the real time saving compounds, because by week four you're editing rather than generating from scratch.

The part most people writing about this skip over

Here's the bit that doesn't get said enough: a first-draft AI lesson plan is, by default, built for an imaginary average student who doesn't exist. If you type "create a lesson plan on fractions for a 10-year-old" and use whatever comes out, you'll get something perfectly competent and completely forgettable, and any parent paying £35 to £50 an hour for private tutoring in the UK will eventually notice their child is getting the same shape of lesson every week regardless of how the previous session went.

Kids notice too, even if they can't name why. A lesson that responds to the mistake they made last Tuesday feels different from one that doesn't, and that difference is most of what parents are paying for when they choose a tutor over a workbook. The tool doesn't do that work for you. You have to bring the specific detail about the specific child every single time, which takes discipline most tutors don't build into their week because it feels like admin rather than teaching. It isn't admin. It's the whole point.

There's also a data point worth being blunt about: several of the "AI education platform" tools ask you to upload real student names, grades, and sometimes school information to personalise output. Under UK GDPR, that's personal data about a minor, and you as the tutor are the data controller, not the software company. Before you upload a class list or a child's full name and school, check the platform's data retention policy and stick to first names or initials wherever the tool allows it. Most tutors never think about this until a parent asks, and by then it's an awkward conversation to have retrospectively rather than a five-minute decision made upfront.

What I'd tell a tutor to buy

If you're tutoring one or two subjects for under ten hours a week, the free tiers of ChatGPT and MagicSchool AI will cover you, and I'd spend zero pounds until you've outgrown them. If you're running a full tutoring business with a website, bookings, and repeat clients, I'd pay for ChatGPT Plus at £16 to £20 a month and pair it with MagicSchool AI's individual Pro tier for the differentiation and rubric tools ChatGPT doesn't do as cleanly. That's under £40 a month combined, against a time saving that, on the numbers above, is worth several hours of billable tutoring time every single week.

If you're building this into an actual business rather than a side hustle, the discipline of running any small enterprise applies here too, and it's worth reading up on how other small businesses are staying current with fast-moving AI tools; my weekly roundups, like this recent AI news update for small business, cover new releases and pricing changes in this space as they happen, since these tools update their features and prices more often than most tutors have time to track.

Building your own AI teaching skills, not just your tools

The tutors getting the most out of this aren't the ones with the fanciest software, they're the ones who've spent an hour or two learning how to prompt well, structure a plan, and edit an AI draft with a teacher's eye rather than accepting the first output. If you want to build that skill rather than picking it up by accident, there are structured options worth the time; I've pulled together a list of remote continuing education resources that, despite the marketing framing, work just as well for tutors wanting to sharpen how they use AI in their own practice.

It's also worth remembering the value that made tutoring work before any of this software existed. AI can generate the plan, but the reason parents hire a human tutor rather than an app in the first place is still the relationship and the judgement, something I've written about at more length when looking at how tutoring builds real skills and knowledge that a static resource never quite manages on its own. The tools should support that relationship, not replace the reason it exists.

Good lesson planning has always rested on a few core teaching ideas, like matching your objective to the level of thinking you want from the student, whether that's simple recall or genuine problem-solving, a distinction education researchers have mapped out for decades in frameworks like Bloom's taxonomy. AI tools are good at generating content against whichever level you ask for. They're not yet good at knowing which level a specific ten-year-old needs today versus what they needed last Tuesday. That judgement call still sits with you.

Related: the ai tools page.

Frequently asked questions

What is the best free AI tool for tutors to create lesson plans?

ChatGPT's free tier is the most flexible starting point for most tutors, handling everything from GCSE maths worked examples to primary reading comprehension questions, though it does hit usage caps during busy weeks. MagicSchool AI's free tier is a close second and includes purpose-built templates for differentiation and rubrics that ChatGPT doesn't offer out of the box.

Can AI tools replace a tutor's own lesson planning judgement?

No, and this matters more than most tool comparisons admit. An AI tool will happily produce a generic, competent lesson plan for an imaginary average student, but it has no idea what a specific child got wrong last week unless you tell it directly. The tutor's job of noticing patterns and adjusting week to week is still entirely human.

Is it safe to use AI tools with real student information?

Be careful here. Under UK GDPR, a tutor uploading a child's full name, school, and grades to a third-party AI platform is acting as a data controller for a minor's personal data. Stick to first names or initials where possible, check each platform's data retention policy, and avoid uploading anything you wouldn't be comfortable explaining to a parent directly.

How much time can AI tools save on lesson planning?

In one real case, a GCSE maths tutor cut her weekly planning time for eight students from about ninety minutes to roughly twenty minutes by using a consistent prompt template built around each student's recent test performance and error patterns. Results vary by subject and how much detail you feed the tool, but a reduction of 60 to 80 percent is realistic once your prompt template is set up.