Nasaq Arabic AI training content with a documented human review before publishing
Every question is tied to a lesson objective, and nothing publishes before a logged human review.
- 58→94
- programme readiness after fixes
- 3
- sample programmes
- 1
- publish gate
- 100%
- questions tied to objectives
The problem
Most AI training-content tools output slides and multiple choice, with no clear link between a question and its objective and no record of who reviewed what.
Overview
An interactive prototype of a product I am building for Arabic training academies. AI builds the material, and human review runs in the same pipeline rather than as a side step.
What I built
- An alignment map binding every assessment item to a learning objective
- A publish gate that blocks while critical findings stay open
- A review log recording who reviewed what, and when
- A branching simulation: the trainee decides and the numbers move, so the same scenario ends differently
The engineering call
I do not pretend the model is infallible. The reviewer signs, the signature is logged, and publishing does not pass without it.
