Generat_Ed lightbulb

Simon's Builder Log

What's Being
Built Next

Who's building this

A teacher who got fed up waiting for someone else to fix it.

I'm Simon. I'm an FE tutor and I built Generat_Ed because nothing else did what I actually needed in a classroom. Most AI lesson tools generate generic slides that look like they've never been near a real student. I wanted something built around retrieval practice, real FE subjects, and the kind of pedagogy that holds up in an observed session.

If you're a teacher reading this — this page is for you, not investors. I'm telling you what I'm working on and why, in the order it makes sense from a classroom outward. If something here sounds useful, or you've got an idea I haven't thought of, the feedback link is at the bottom. I read every one.

Latest thinking

Coming next

Generat_Ed Primary — built for the classroom downstairs.

Secondary and FE teachers will know this conversation. A primary colleague asks if they can use the tool. You show them a generated lesson. You both see the problem immediately. The cinematic design isn't wrong for secondary. For a Year 2 class on an ageing projector, it's a different planet.

Primary is a genuinely different job. Not simpler — differently complex. I've spent real time on this, and the primary version needs to be built from the ground up, not adapted from a secondary template.

Every maths lesson will align to White Rose Maths Version 3 — the scheme used in 70% of UK primary schools. The tool will know which representations belong with which year group and topic. Bar models, ten frames, part-whole diagrams, place value charts — not just images, but the pedagogically correct images for each stage of the Concrete-Pictorial-Abstract journey. A maths lesson that skips to equations for a Year 2 class isn't just wrong — it's the kind of thing that makes primary teachers distrust a tool permanently. The Flashback 4 retrieval starter is built in as standard.

KS1 slides are built to a maximum of eight words of instructional text. Everything else is visual. Clean backgrounds, subject-coded by colour, checked against the back-row test — if a child at the back of a standard primary classroom can't engage without moving forward, the design fails. For English, every text shown to KS1 children carries a phonics phase flag. The tool won't generate a Phase 3 lesson with Phase 5 words hiding in a caption without flagging it for explicit pre-teaching.

The You Do slide carries three access variants for the same task. Not separate worksheets that mark children out in front of their peers — one slide, three entry points. The teacher sees all three. The children see one. Oracy is structural, not a suggestion — sentence stems matched to the subject and year group appear on the slide at the moment they're needed. SEND considerations are embedded from the start, not added as a note at the bottom.

I haven't built this yet. The FE and secondary version needs to be solid and trusted first — that's where the focus is right now. But the architecture is done, the curriculum data structures are designed, and the visual rules are settled. When primary launches, it'll be built properly. Not this year. But properly.

In planning

June 2026

The Vault is open. What comes next.

The Knowledge Bank is built and live. Every lesson you generate now deposits its learning outcomes into The Vault — your own growing archive of what you've taught, organised by unit and awarding body. The Revision Wheel is in there too: spin it, land on an outcome, a pre-generated question appears, teams score. It works, and it's already more engaging than anything I expected when I started building it.

Now I'm asking a different question: what else can you do with a bank of outcomes? The spinning wheel is one format — but I want to build more. Different revision structures that suit different moments in the term: end-of-unit, mock prep, low-stakes retrieval on a Friday. I'm sketching out new game formats and I'll build the ones that feel genuinely useful in a real classroom, not just clever on a screen.

I'm also building out the slide bank — a library of activity templates the tool can draw from when generating lessons. More variety in how a topic gets taught. More ways for students to engage with content beyond reading and answering. The goal is that two lessons on similar topics feel genuinely different, and that every generation gives you something worth using.

And the community piece — I'm more excited about this than anything else right now. Teachers sharing what works. The best ideas built into the tool. Real classrooms making the product better. That's coming once the first users are in and I know what's actually needed. Not long now.

Live + in progress
Earlier entries

After you've generated a few lessons on the same unit, you've already built up a bank of learning objectives. I want to put those to work in a proper revision session — not just another lesson, but a game.

I'm thinking a team tiles board: outcomes hidden under coloured tiles worth different points, teams take turns picking, a question gets revealed, correct answer wins the points. The kind of thing that works at the end of a unit when you want students competing but actually recalling content. No names on the tool — students get numbers, teacher keeps the seating plan on paper. Impartial, GDPR-clean, and it removes the awkwardness of the teacher always picking the same people to answer.

A spinning wheel version is coming too for quicker check-ins. This one's taking time to get right — I want the questions to be genuinely diagnostic, not just fun.

✓ Built

A lesson for a Level 3 apprentice on a Wednesday afternoon is not the same as a CPD session for qualified professionals on a Friday morning. The content might overlap. The language, the register, the assumptions about prior knowledge — none of it should.

Right now, Generat_Ed knows your subject and your level. What I'm building toward is something deeper: slides that understand the context of the learning. CPD sessions for adult professionals use a different vocabulary — reflective practice, continuing development hours, workplace application, professional standards. A session on COSHH for a qualified technician looks nothing like a COSHH intro for a Level 1 student, and it shouldn't sound like it either.

I'm working on topic-specific slide packs that carry the right language, the right frameworks, and the right professional register baked in from the start. Not templates you edit — slides that already speak the language of the room before you've touched them.

In research

Right now when you hit Generate it goes quiet for 30–40 seconds. The lesson is building the whole time — Claude is working through the structure, the slide types, the questions — but you can't see any of it happening.

I want to change that so you can watch it build in real time. Each section appearing as it's written, like watching the lesson come together. It'll feel faster even when it isn't, and it'll give you a better sense of what the AI is actually doing with your topic.

Planned

The generated lesson is a starting point, not a finished product. Sometimes you want to drop a slide, move the discussion earlier, or add a retrieval starter on a specific topic you covered last week.

I'm building a proper slide editor — drag to reorder, delete what you don't need, and a mini-generate button that creates a single new slide from a short description you write. It'll slot straight into the existing lesson and pick up your images automatically.

In development

Teachers are the best source of teaching ideas. Not AI, not consultants — teachers who've stood in front of a difficult group on a Friday afternoon and found something that actually worked.

I want to build a way for tutors using Generat_Ed to share those moments. Not a forum, not a rating system — something simpler. If you've run a lesson activity that landed brilliantly, you describe it: the slide type, the subject, what made it effective. The best ones get reviewed and built into the tool as new slide formats, available to everyone. Tutors who contribute get credited with free lessons. The tool gets better. The profession gets better.

I'm not building this until the first users are in and settled. But it's the feature I'm most excited about long term. If you want to be part of this when it opens — reply to any email from Generat_Ed with the word "Community" and I'll make sure you're first in.

After first users

The tool already generates different slide types depending on your topic, subject, and duration. But I want to push the variety further so two lessons on similar topics feel genuinely different — different question styles, different visual layouts, different ways of structuring the reveal.

This one's not a single feature — it's something I'm improving with every batch of lessons that goes through the tool. If you generate a lesson and think "this would work better if..." — hit the feedback link and tell me. That's genuinely how this gets better.

Continuous

Getting students interested is genuinely the hardest thing in teaching right now. They've got TikTok, Instagram, YouTube — all of it designed by billion-dollar companies specifically to be more compelling than anything a teacher can put on a whiteboard. That's not a fair fight, and pretending it is doesn't help anyone.

What I can do is make the lesson itself feel worth being in. Generat_Ed already builds in sounds, confetti, challenge moments, team competitions, and visual reveals — small hits of engagement that reward attention rather than punishing distraction. I want to push this further. Not gimmicks — purposeful moments that land at the right point in the lesson when energy dips and you need to pull the room back.

The goal is a lesson that feels like something worth paying attention to — on its own terms, not because someone told them to. That's what Generat_Ed is really for. Not to replace what you do — to make what you do worth watching.

Continuous