Qubits AI

From feature-rich to teacher-first.

Role
Product designer
Team
Product manager, 2 designers, development team & CTO
Responsibilities
Product strategy, user research, UX/UI, AI experience design, prototyping, design systems

with National Geographic Learning

The problem

Qubits AI is an AI learning management system for teachers, students and district school admins across 13 countries, primarily the US, UAE and India, built in partnership with National Geographic Learning.

We kept adding value. But something broke.

A platform packed with powerful tools that teachers simply weren’t using. The more capabilities we shipped, the more friction we inadvertently introduced.

“I know the feature exists. I just can’t find it.”
Teacher
Too many features. Too many places. Too much friction.

The platform before

Every new capability lived in its own place.

AI Live Classes, Content Creation, AI Assistance, Teacher Resources, Analytics and Lesson plan generation. The win was real. The cracks were forming.

Research

So I proposed co-creation.

Inspired by the IBM user sponsors program: stop building, start partnering. I started asking how we could make teachers use the existing features better.

I prioritised co-creation workshops, along with feature development, mapping teacher workflows from initial lesson planning to live instruction.

Co-creation workshops
2
Teachers partnered in depth
9

An MIT study of 300+ enterprise AI deployments found that 95% of generative-AI pilots delivered no measurable business impact. The 5% that worked built AI into the workflow, backed every claim with evidence, and kept the human as the author. Source: MIT NANDA, “The GenAI Divide: State of AI in Business 2025”.

The real core issue

The problem wasn’t the number of tools or features. It was the lack of context.

Capabilities were completely isolated. No shared memory. No continuity between different steps of a teacher’s day.

Before · feature-centric

  1. Find a tool
  2. Enter context
  3. Generate
  4. Move somewhere else
  5. Repeat

After · context-centric

  1. Teacher intent
  2. Context
  3. AI
  4. Recommendation
  5. Action
  6. Outcome
  7. Context continues

The design

The design concept

A teacher co-pilot.

I came up with this concept. The teacher shouldn’t have to know which specific feature they need. They should simply say what they want to accomplish.

By keeping the educator in the pilot seat while AI manages the data connections, we shift the experience from technical configuration to pure workflow enablement.

The AI ecosystem

We don’t need a play store of AI features.

Our ecosystem acts as a single, cohesive engine. Multiple context streams flow into the Qubits AI engine to surface relevant, proactive teacher recommendations.

The designs

Three teacher stages. One connected system.

01 · Plan

Planning before the class.

One AI workspace pulls student data, schedules and resources together, so planning starts from context instead of a blank tool.

Before

AI was added to every tool separately, with no context. It overwhelmed teachers and made new features hard to adopt.

After

One space that brings context from across the platform, without navigating away or losing focus on the goal.
  1. Context comes to the plan. Recent assignment analytics for the class, without leaving the plan.
  2. Starts from what students struggled with. The plan opens with the reason, then what the AI prepared.
  3. One step to act. The plan goes straight into the class flow.

02 · Teach

Intelligence in the moment.

A contextual recommendation appears inside the teaching experience: live feedback, instant quiz generation and assigning. No tab switching, no searching.

03 · Analyse & evaluate

Turn data into a decision.

Data that used to sit on dashboards as numbers is now turned into action items and recommendations teachers can work on immediately.

Before

Blocks and numbers. Insights existed, but were buried in a sea of numbers.

After

Goals and recommendations in one space, ready to act on.
  1. Today’s goals. Each task with its progress, so teachers see what’s left today.
  2. Teacher recommendations. Suggested next steps, each with its own action.

InformationIntelligenceAction

Impact

What changed after launch.

Sales growth
40%
Sales data, 2024–2026.
Feature adoption
51%
Up from 20%. Teacher survey and business reports.
Increase in user satisfaction
70%
Survey of the 30 teachers in the co-creation programme.
Teachers spent less time preparing
21 of 30
70% of those surveyed.

Takeaway

The problem wasn’t that we built too much. It was that we connected too little.

The next evolution of AI wasn’t about adding more capabilities. It was about giving those capabilities shared context.

Contact

Let's work together to create something amazing. I'm always excited to discuss new opportunities and innovative projects.

Get in Touch View Resume