Solution
In the ed-tech space, scalability always posed a significant challenge. Newton School's mission is to provide learning solutions at scale, but technological limitations constrained us until the release of ChatGPT. AI introduced new possibilities, allowing us to offer students immediate feedback when integrated into their workflows.
We developed an AI-powered online tool called "AI Tryouts" for tech interview practice, featuring a repository of questions asked in various companies. Users can verbally answer these questions and receive instant feedback. The tool evaluates their responses based on multiple parameters and provides feedback on their answer delivery along with suggestions for improvement.
User Flow
Users arrive at Tryouts through various channels, including ads and different mediums. Consequently, distinct entry flows are designed to cater to the diverse ways users access the platform.
User Flow
img
Landing Page
I opted for a 2-column layout for the landing page of Tryouts, catering to both signed-in and non-signed-in users.
On the left, the hero section provides a brief description of the product offering, while on the right, users can find the selectable modules to practice questions.
This layout was chosen as our target audience typically doesn’t scroll through the product extensively. Placing actionable elements in the first fold minimizes friction, making it easier for users to engage with the product.
Tryouts Landing Page
img
Module Page
The module page features a list of questions within the module, providing users with the option to search for specific questions and filter them based on the topics they want to practice.
module page
img
Question Page - Answering a Question
Once the user selects a question, they are prompted to provide a verbal answer and grant microphone permissions for Tryouts. After recording their response, a transcript is generated. Users have the ability to make minor edits to the transcript and re-record their answer if they are unsatisfied.
Upon submission, the AI processes the response and provides feedback.
You can experience a demo yourself by following the link provided just below the case study title or take a look at the video below.
Solving a Question - AI Tryouts
VId
Impact
Following its launch in June 2023, within four months over 28,000 users signed up on the product. Got over 6,000 unique question submissions, with over 200 users completing all the questions and earning certificates.
Scope for Future Versions
Currently, users are required to navigate back to the module page to select another question. However, we aim to enhance the user experience by enabling seamless navigation of questions and modules directly within the question page. This improvement is anticipated to further reduce friction in the user journey.
Users currently lack visibility into their previous attempts for a question. We plan to introduce a repository within the question attempts section, enabling users to review their previous answers and learn from their mistakes.
Looking ahead, our vision includes allowing users to select topics and difficulty levels, with a set of chosen questions grouped and presented as an interview.
As AI capabilities continue to advance, the future vision is to create an experience similar to an online interview, where users engage in conversational interaction with an AI interviewer, eliminating the need for manual question selection. (This idea is still very far fetched)
Final Thoughts
Navigating tight timelines presented a unique challenge for this project. Collaborating closely with my Product Manager, we aimed to deliver Tryouts within an ambitious 2–3 day window. The collaboration involved scoping out some features to effectively ship the product while leaving room for future enhancements, all while ensuring the overall integrity of user flows.
Experiencing the adrenaline rush of shipping within such a crisp deadline was a novel challenge for me. It highlighted the importance of efficient collaboration, prioritization, and making strategic decisions to meet project goals.
Thank you for reading this far✌🏻.
