AI in Education

AI-Powered MusicTechnologyEducation

How one A-Level teacher built a complete learning platform using AI tools — no engineering team, no funding, no code experience

Scroll to explore

AI in Education

01

AI-Powered Marking

Teachers spend evenings and weekends marking papers that could be assessed in minutes. AI marks multiple-choice, calculations, and written responses against official mark schemes — giving students immediate directive feedback (Hattie) instead of a five-day wait that delays the corrective cycle identified by Wiliam & Black.

BeforeA weekend lost to marking 30 papers
After16 students marked in one afternoon
Live Demo
02

Blind Marking

Bloom’s Two Sigma research found teachers treat students unequally, with some receiving encouragement while others are largely ignored. Anonymous student IDs remove that bias entirely — marks are finalised before identities are revealed. Students noticed the difference, reporting that their marks felt fairer across the board.

BeforeRecognise handwriting, adjust expectations
AfterMark the work, not the student
Live Demo
03

Live Data Sync

Most departments keep grades in three or more disconnected systems — spreadsheets, portals, mark books. Bloom identified that responsive teaching requires immediate access to performance data. This two-way sync means grades entered anywhere appear everywhere automatically, enabling the data-driven decisions that mastery learning depends on.

BeforeCopy marks between 3 different systems
AfterEnter once, available everywhere
Live Demo
04

Interactive Tools

Craik & Lockhart’s Levels of Processing research shows that how material is processed matters more than the intention to learn. These 14 interactive tools force deep semantic processing — students build synth patches, shape EQ curves, and hear compression in real time, combining Paivio’s dual coding (visual + auditory) with hands-on application.

BeforeWatch a video, hope they remember
AfterHear it, shape it, understand it
Live Demo
05

Revision Hub

Dunlosky et al. rated practice testing and distributed practice as the two highest-utility learning techniques. AI-marked quizzes across 6 topics implement both — retrieval practice with spaced intervals, showing students exactly where their gaps are. As Rosenshine prescribed: begin each session with review of what came before.

BeforeRevise everything and hope for the best
AfterFocus on what you actually need to learn
Live Demo

The Numbers

80%
Less Marking Time
5→0
Day Feedback Wait
14
Interactive Tools
0
Bias in Marking
1
Teacher. No Dev Team

The Approach

Built with AI, not an engineering team

Built using Claude Code — a conversational AI coding assistant. You describe what the tool should do in plain English, and it writes production-ready code

Rapid iteration with real students — build a prototype in the morning, test it in the afternoon lesson, refine based on what actually happened in the room

No engineering team, no budget, no procurement — just one classroom teacher who understands the problems students face every day

Domain expertise is the superpower — AI handles the code, but knowing which pedagogical problems to solve and how students learn is what makes the tools effective

The Bigger Picture

What this means for schools

Scalability

Serve an entire department, not just one teacher — the same platform handles any subject with exam-style questions and mark schemes

Cost

Less than a single textbook set per year — no expensive licenses, no per-seat fees, no procurement process

Data-Driven Insights

Real-time visibility for everyone — teachers, students, parents and leadership see progress instantly, not at parents' evening

Fairness & Equity

Blind marking removes unconscious bias — consistent AI feedback means every student gets the same quality, regardless of who marks their work

Get in Touch

Interested in what AI
can do for your school?

Whether it's a half-day workshop, a department pilot, or just a conversation about where to start

LinkedIn

About

Mike Lehnert

Music technology teacher. AI tool builder. Helping educators figure out what actually works.

After 20 years in the classroom — in schools, at Leeds Conservatoire, as a Pearson examiner — I started using generative AI seriously about three years ago. More recently, I've been building tools with Claude Code, Obsidian, Gemini, and NotebookLM. My approach is to use multiple AI systems rather than locking into one — different tools for different jobs.

I've built an Obsidian vault system for organising teaching materials, developed grading and feedback tools, and created interactive resources for A-Level Music Technology. I've documented this work on YouTube to help other educators explore what's possible.

What I've found is that AI works best when it walks alongside you as a teaching assistant — not doing the work for you, but helping with the repetitive stuff so you can focus on actual teaching.

Schools need help figuring this out. Most AI advice is either too abstract or comes from people who haven't been in a classroom recently. I'm still teaching daily, so I know what actually works and what doesn't.

MIE ExpertApple Certified TeacherPearson Examiner20 Years Teaching

Open to consulting, training, and collaboration with schools ready to integrate AI properly.

Connect on LinkedIn