AI Quiz

What is it?

AI Quiz is an AI-powered feature inside PhysicsWallah that lets students generate personalised quizzes instantly from an image, a YouTube link, a PDF or any study material. The goal was to turn quiz creation from a slow, high-friction process into something that feels instant and habit-forming, especially for first-time users who drop off before they ever see the value.

The challenge

PhysicsWallah’s AI Guru tools had strong adoption (roughly 60–70% of subscribed users asked at least one query), but quiz generation was slow, so first-time users left before experiencing value. We needed to deliver an immediate aha moment, keep returning users engaged, and strip the friction out of quiz creation: turn a utility into a habit.

The approach

I started by studying what works in products like Duolingo, Quizlet, Brilliant and Solvely, the ones that make learning feel instant and rewarding. Five patterns emerged:

• Never overwhelm new users.
• Begin with purpose, not features.
• Progressive disclosure.
• Simplification.
• Smart nudges.

From there I formed three hypotheses around student-created quizzes (improving engagement, converting free users, and impacting retention) and mapped two real risks: latency killing the first run, and quizzes not feeling personal enough.

Competitive analysis.

Hypothesis & risks

• Letting students build quizzes from their own material lifts engagement and retention.
• It gives students who don’t pay an easy way to try the AI tools.
• Students create quizzes regardless of location, curriculum or stage of prep, which feeds retention.
• Risk: latency on the first run drives drop-off; quizzes that don’t feel personal blunt engagement.

Hypothesis & risks framework.

Concepts explored

I explored multiple concepts for both the first-time experience and for friction reduction, each optimising for a different behavioural hook. Two experiences to solve for:

• The first-time experience.
• Reducing friction, with an exit strategy.

Concepts explored.

Decisions

Stakeholders wanted the best of each concept, so the final direction held on to:

• An emotionally safe entry point.
• A quiz with easy wins: the “I did it” moment.
• Achievement leading into exploration.
• A calm latency screen.
• Subtle feature education.
• A proper exit strategy.

Final design