Designing Educational AI Features for Brainly

Brainly, to maintain its position as a leading ed-tech company, must continuously update its value proposition with the newest tech, which now includes generative AI. This innovation brings forth unexplored UX problems and challenges related to strategy, tech limitations, or communication clarity. Below is a summary of early AI-based implementations.

Company

Brainly

Years

2022 - Ongoing

Categories

UI and UX Design
Leadership and Strategy
Visual Design

My role

Both hands-on work and leading projects related to the topic. Leading and managing design processes, research, and stakeholder communication.

Background

The ed-tech market is one of the most severely impacted by "The AI Revolution" due to a growing interest among younger demographics for innovative learning features. I've worked on several projects to address this opportunity, each presenting distinct challenges, ranging from communication methods to working within specific technical limitations.

Communicating AI features is particularly difficult due to their novel technical nature, with Brainly's younger user base often not fully comprehending their value proposition.

Multiple technical unknowns and constraints had to be solved – while generative AI is amazing tech-wise, some design decisions were made around dev-related constraints.

AI opens up unique reputational risks related to the perceived quality of the available content (users’ primary concern) due to its generative nature.

The Process

Understanding the overarching business context is crucial in delivering value that benefits both the user and the bottom line. Stakeholder interviews were a crucial part of all design projects related to generative AI as the technology has important ramifications for ed-tech.

Evaluating the technical constraints in light of strategic priorities. After understanding the end-goals of potential improvements, I could start working with developers on establishing technological constraints of generative AI. These would be used as design guardrails.

Desk research, product audit, and competitive analysis. As use of generative AI in products is still largely in infancy, I needed to familiarize myself with common patterns emerging on the market. Through this exploration, I could establish unique solutions that still follow early standards.

Strategic thinking and structuring the experiment rollout. Many of the proposed changes and designs were planned ahead to ensure the right opportunities are verified with hypothetical solutions.

Wireframing – tech assessment, testing, and stakeholder feedback. Creating simplified low-fidelity designs followed all research and discussions. I could use these “blueprints” to swiftly align with stakeholders and developers.

Designing the final product with cross-team consultation. While the exact details depended on project, all of them were created under the same scrutiny and followed best possible practices – RWD, accessibility, design system alignment, among other aspects.

Handover and launch. Learning from experiment results. “Continuous growth” was the guiding approach for anything AI-related, as its novelty inherently leaves room for improvement. Quantitative A/B testing helped validate specific hypotheses and quantify opportunities' potential.

I've illustrated the AI stand-in character (Ginny) to guide users through complex back-end processes. As the technology is challenging to grasp for the general public, it's essential to make it more approachable, fun, and conversational.

Qualitative research proved users aren't generally familiar with AI specifics. Further testing and interviews informed various minute aspects of the UX, from copywriting to feature placements.

Due to AI's novelty, it was essential to establish a factual basis for product decisions. I've maintained documentation compiling competitive analyses and the results of early internal experiments.

The Outcomes

Early implementations of generative AI explored possible avenues for adding the tech in the already-established user flows within the product. Concepts and business hypotheses validated through AB-testing clarified the strategic and UX direction of further refinements related to generative AI.

The first few experimental concepts established product-wide solutions for communicating AI; eventually, they'd be added to the design system and brand identity.

Validating hypotheses and possible strategies through A/B experimenting was critical in shaping the strategic direction of the product's generative-AI implementations.

As a designer versed in the topic, I guided others by providing feedback on the user experience of their own AI-based design projects.

The Question-and-Answer Page (the most visited part of the product) was fundamentally redesigned to better support new AI-based features and improve the overall UX without confusing returning users.

First exploratory implementations of AI-based features worked within predetermined contexts: simplifying and expanding user-made content.

Visuals and animations clarified features that could be difficult to grasp without sufficient technical knowledge.

The embeddable widget reuses an existing AI-chat technology. This cost-effective solution improves Brainly's unique standing in the ed-tech market.

The widget's entry point is customizable to satisfy potential partners' broad spectrum of needs.

Predefined shortcuts can be used to instantly personalize learning content in the AI chat, significantly increasing accessibility.

I've represented the AI mascot (Ginny) in a variety of scenarios to visualize intricate technical parts of the product. As AI is still largely a novelty not comprehended well by the general public, even some of its least complicated aspects needed proper contextualization.

Showcase 1

Designing an End-to-End Monetization Strategy

See Full Case Study

Leading the design of the entire monetization model for Brainly, the ed-tech leader

Facilitating cooperation between multiple teams and stakeholders

Product strategy for continuous growth, routine testing to validate business concepts

Interactions driven by user research and data analytics

As usual, there is still so much more to cover but only so little time. Feel free to check out my social media links or drop me a message to learn more about my work, skills or availability.