AI Revision Bot gives every student instant, personalised feedback on their written answers, powered by Google Gemini. Teachers get a full marking dashboard, analytics, and a badge reward system. No subscription. No per-student fee.
What's included
Built by a teacher, for teachers. Every feature came out of real classroom use at Exeter College.
Students submit written answers and receive structured, mark-scheme-aware feedback within seconds via Google Gemini 2.5 Flash. No waiting for a teacher to mark a pile of books.
Teachers review AI feedback and apply Red, Amber, or Green ratings. The system tracks performance over time and shows clear improvement trends per student and per class.
Students earn badges across four tracks: Green %, Amber/Green %, No-Red streak, and Green streak. Badges show in the nav bar and on a dedicated progress page, and they are genuinely motivating.
Students can upload diagrams, sketches, or screenshots alongside their written answer. Gemini analyses both text and image together, ideal for science, maths, and computing.
A live admin view shows all student responses, uploaded graphics, and teacher notes. Auto-refreshes every 30 or 60 seconds so you can monitor the room without touching your keyboard.
Questions, mark schemes, and student answers all support bold, italic, subscript, superscript, code blocks, and lists via a TipTap editor. Write H₂O and x² properly at last.
Time-based RAG progress graphs, class comparison tables, per-subject breakdowns, and CSV export for your own records. AI-suggested ratings fill in the gaps for unreviewed responses.
Bulk CSV upload for your entire cohort, automatic welcome emails, forced password change, account deactivation, and class management. Everything a department head needs.
All API endpoints require signed bearer tokens. Admin functions are locked behind role-based access control, and directory browsing is blocked. Student data stays on your own server, or ring-fenced on ours, and is never shared with third parties.
The RAG system
The Red/Amber/Green rating system gives students a clear signal without a number grade to argue about. Patterns in the data tell you who needs help before they fall behind.
When a teacher (or the AI as a fallback) rates a response, students see their running totals on their progress page:
The analytics dashboard uses COALESCE(teacher_rating, estimated_grade) so class statistics remain meaningful even before teachers have reviewed every answer.
Reward system
Hand-crafted steampunk medallions reward consistent progress across four tracks. They appear in the menu bar and on the My Progress page, turning revision into something worth coming back to.
Earned as a student's share of Green-rated answers climbs, from 25% all the way to 90%.
Rewards overall security: the combined proportion of answers rated Amber or Green.
For consecutive Green answers, recognising a run of strong, secure responses.
Celebrates consistency: a run of questions with no Red rating in sight.
How it works
Teachers build the question bank. Students practise. The AI does the marking. You review, rate, and act on the data.
Upload questions and mark schemes for any subject you teach. Rich text supports subscripts, superscripts, code, and lists right out of the box.
Questions are served in random order. Students type their answer, or upload a diagram, and submit. Answered questions get a green tick so they can track their coverage.
Google's Gemini 2.5 Flash reads the answer against the mark scheme and returns structured, constructive feedback within a few seconds.
Your teacher dashboard shows every response. Apply a RAG rating and add a comment where needed. The analytics update immediately.
Badges earn as performance improves. Progress graphs show RAG trends over time. Students who are flagging are easy to spot before it matters in an exam.
Subject coverage
If you can write a mark scheme, AI Revision Bot can assess against it. The question bank is yours to build, covering exactly the topics your cohort needs.
Currently in use at Exeter College for:
Technology & setup
AI Revision Bot runs on a standard PHP/MySQL web host, the kind your school may already have. There is no SaaS dependency, no ongoing vendor relationship, and no lock-in.
Each school provides and pays for its own key, and the cost is tiny. Visit Google AI Studio, sign in with a school Google account, and create a key. You then add it through your own admin account, where it is stored encrypted in the database, never in a plain config file and never shared with us. This works the same way whether you self-host or we host for you. Google's free tier comfortably covers normal classroom use; beyond that you only pay for what you use, typically around 60p a month.
An enquiry is free (as is the aibot)and takes five minutes. We send you access to the codebase, a setup guide, and the SQL database template. All we ask is that you let us know how it goes - because I'd like to publish some research and case studies.
Register your school — email us using the button below and we'll be in touch with download links and setup instructions.
Get your Gemini key — takes two minutes at Google AI Studio. You add it in your admin account, where it is stored encrypted. Free to start.
Deploy it yourself — copy the files, import the SQL template, add your config, and you're live. The GitHub README walks you through each step.
Or let us host it — we'll run your instance on Exeter College's servers, with your data ring-fenced from every other school and never shared with third parties. Because we use the platform ourselves, it already meets our GDPR and safeguarding standards — and it's still completely free.
Build your question bank — add subjects, topics, and questions. Bulk CSV upload lets you get a whole year group set up in minutes.
Your details are emailed straight to the AI Revision Bot team. Nothing is stored, and nothing is shared with anyone else.
