Marvely

Marvely is an AI-powered speaking tutor designed for language learning, especially in high-stakes exam settings where practice quality and feedback quality matter a great deal.

Students work through realistic speaking scenarios, receive immediate rubric-aligned feedback, and can try again straight away while the context is still fresh.

It also keeps teachers in the loop through transcripts, cohort insight, and tools that turn student performance into follow-up teaching actions.

What makes Marvely compelling is that it is not just a generic AI interface with educational language wrapped around it. The product is built around a tighter learning loop: attempt, feedback, retry, and improvement.

Its strength is the combination of AI capability with careful product structure, exam alignment, and teacher-facing workflow design.

That makes it feel less like novelty and more like a serious learning product designed to fit real classrooms, real curricula, and real constraints.

Marvely began as an idea from one of my engineers, who built the first prototype in a weekend.

I worked with that engineer and our designer to turn the prototype into a production-ready product.

In the last six months I have personally taken over the coding whilst we iterate toward product-market fit.

I am particularly focussed on optimising funnel conversion and the user journey through activation, retention, revenue, and referral, and I personally run the paid advertising campaigns whilst we iterate toward our target CAC and a strong LTV.