OverviewProblem overviewResearch and strategyDesign approachUI design and implementationFinal designsImpact and learnings

Astra AI - Web app design

My role

Product Designer -  Research, Graphic design, UX & UI Design

Team

Klemen Selakovič, Product lead

Me, Product Design

Timeline & Status

4 Months, Live

Overview

Astra AI is an AI-powered math tutoring platform designed to provide personalized, step-by-step solutions for students in elementary and high school, as well as anyone looking to improve their math skills. As the company moved beyond its MVP stage, they needed a more refined product that could scale while maintaining a seamless user experience.

I was contracted as a product designer to improve discoverability, enhance usability, and implement new features that would support Astra AI’s evolving needs.

PROBLEM OVERVIEW

Astra AI was growing beyond its MVP and needed a more refined experience

Identified problem statements

Scaling beyond the MVP – As Astra AI moved past its initial launch, it needed a more polished and structured product without losing agility.

Enhancing usability for learning – The existing tools, like the math keyboard, onboarding flow, and tutoring options, needed updates.

Expanding the feature set – The core product was simple, but adding new functionality became a challenge, requiring thoughtful integration.

Introducing customization – New settings and chat history features were needed to give users more control over their learning experience.

RESEARCH AND STRATEGY

Understanding the product’s evolution

Navigating a fast-paced startup environment

My work was largely independent, with tasks assigned directly by the Product lead. While I wasn’t able to have a direct contact with users, I based my decisions on available data and market trends to ensure practical and effective solutions.

Strategy for coming up with solutions

Since Astra AI was moving beyond its MVP phase, I had access to a few previous user testing reports but primarily relied on competitor research and industry best practices. However, given the emerging nature of AI-powered tutoring platforms, there were times when I had to improvise and develop my own solutions.

Working with the current product

To maintain consistency, I adhered to Astra AI’s existing design language. The objective wasn’t to rebrand the product but to refine and expand upon what was already in place, ensuring a seamless experience for existing users while integrating new features smoothly.

DESIGN APPROACH

Finding the right approach in a fast-moving environment

Prioritizing the right tasks

Each week, the Product lead and I reviewed tasks to determine priorities based on implementation difficulty and impact on the product. High-impact tasks often required more in-depth exploration, while simpler ones needed quick, efficient solutions.

Coming up with solutions

Given the rapid pace of development, instructions were often flexible, requiring me to generate multiple design solutions. I prepared several variations for each feature, presenting them to the Product lead for evaluation. The selected approach was then refined and implemented into the final product.

1.0 Solution 1.

Image

1.1 Solution 2.

Image

1.2 Solution 3.

Image

UI DESIGN AND IMPLEMENTATION

Creating a seamless and scalable experience

Understanding existing design patterns

Since Astra AI lacked formal documentation on design guidelines, I spent time exploring the existing product to understand its structure and design patterns.

By studying how previous elements were implemented, I ensured consistency in new designs. When introducing new patterns, I carefully evaluated their fit within the existing system and how they would scale over time.

Adding a new tutor

One of the major additions to the platform was support for a new tutoring subject: English. This required structuring different language levels and tailoring the user experience accordingly. For beginners (A1 level), instructions were provided in their native language, while more advanced learners received instructions in English.

Additionally, I designed a solution that eased the users into using the new model by showcasing multiple prompt examples, helping users understand what the AI could assist them with.

2.0 English tutor on desktop.

Image

2.1 English tutor on mobile.

Image

Re-designing some of the existing features

Although the direct impact of these changes wasn’t formally measured, reworking some of the core flows significantly improved the product’s usability. The redesigned features made the experience feel more structured and intuitive, allowing users to navigate the platform with greater ease.

3.0 New onboarding.

Image

3.1 Re-designed settings.

Image

3.2 New scientific keyboards.

Image

FINAL DESIGNS

A seamless experience

4.0 Conversation history.

Image

4.1 Archived conversation.

Image

4.2 Updated graphics.

Image

4.3 New pitchdeck.

Image

IMPACT AND LEARNINGS

Evaluating long-term value and key takeaways

Results

Features that stood the test of time – The solutions I designed were successfully integrated into Astra AI and continue to support its growth and evolving needs.

Long-term impact – While no formal analytics were conducted, the adoption of these features highlights their effectiveness in improving the user experience and platform functionality.

Learnings

The power of clear communication – Working in a fast-paced environment showed the importance of alignment between teams, ensuring efficiency and a healthy work dynamic.

Balancing speed with quality – Delivering thoughtful solutions under time constraints was a crucial skill in an industry that is constantly evolving, such as AI-powered education.