AI Builders Lab
Students design, build and pitch a working AI chatbot in one three-hour, in-person session. Currently in pilot with a partner school in Hyderabad.
Three phases, one working prototype
| Phase | Time | Focus |
|---|---|---|
| 1 | 0:00 to 0:55 | Foundations and prompting. How language models actually work, why they make things up, and how to write a prompt that survives contact with one. |
| 2 | 0:55 to 2:05 | The build sprint. Teams start from working starter code and build a chatbot in the browser, connected to a live model through temporary, monitored API routing. |
| 3 | 2:05 to 3:00 | Polish and showcase. Teams customise their bot, then pitch the working thing to a faculty panel. Certificates at the close. |
Division of responsibility
| Assiz Ltd provides | School provides |
|---|---|
| Lead facilitator, technical support, full curriculum | Hall or classroom with projector, Wi-Fi and power |
| Starter code, worksheets and secure API infrastructure | Up to 40 students with charged laptops |
| Printed certificates, judging rubrics, consent forms | 1 teacher coordinator; 2 to 3 faculty judges for the final 30 minutes |
Safeguarding & data privacy
No data is collected
We never ask a student for an account, an email, a phone number or any other personal detail. There is nothing to store, so we store nothing.
Traffic is routed securely
Every AI request passes through a monitored Assiz proxy on short-lived tokens that are revoked when the session ends. No student ever touches a public AI API directly.
Staff are always supervised
Assiz staff work in sight of school faculty at all times and never one-to-one with a student. Content guardrails stay on throughout, and photography happens only where the school holds signed parental consent.
Something they made, and can explain
Most sixteen-year-olds have used a chatbot. Very few have built one, watched it fail, and worked out why. A student who has done that can hold a conversation about how the thing works, which is worth considerably more in an interview than a certificate saying they attended. They leave with a working prototype, the code behind it, and the winning team's bot hosted live for three weeks.
Phase I teaches students to use AI. Phase II teaches them to architect it.
A follow-up session for senior Computer Science cohorts, aimed at the students who finish Phase I early and want to know what sits underneath.
It is a line-by-line teardown of MicroGPT: roughly 250 lines of code that produce an autonomous AI loop. Small enough to read in an afternoon, complete enough that nothing gets hand-waved.
Ask about Phase IIWhat it covers
Agent loops. Why a model that can call itself behaves nothing like one that answers a question once.
Multi-API chaining. What happens when the output of one model becomes the input to another system.
Vector memory. How a system recalls anything at all beyond what is in front of it right now.
Follows Phase I · Roughly 250 lines of source, read end to end · Scheduled separately
Founding pilot rate
£650
Covers facilitation, every student material, the secure infrastructure, printed collateral, and three weeks of live hosting for the winning team's chatbot. Our standard rate is £750 to £950.
Request the pilot