PitchPerfect

Inspiration

We thought back to our own work experiences, especially those tough, sometimes uncomfortable conversations with managers, teammates, or higher-ups. Those moments can be hard to prepare for, and there aren’t many tools that help you get better at them. That got us thinking: there are tons of AI tools that help you prepare for the job, like mock interview generators, but what about preparing for situations in the job? That’s where PitchPerfect comes in.

What it does

PitchPerfect helps professionals practice and improve their communication skills in tricky workplace scenarios. Users answer a quick questionnaire about their current work situation and their boss’s personality type. Based on their answers, the app generates a realistic mock scenario. Users then write or speak their response, and the AI provides thoughtful feedback on their tone, clarity, and word choice to help them improve.

How we built it

We built PitchPerfect with a Flask backend and a React frontend written in TypeScript. The backend uses the Gemini API for prompt generation and critique, Auth0 for authentication, and MongoDB to manage user data and responses. On the frontend, we focused on a smooth, conversational UI that makes practicing scenarios feel approachable and intuitive.

Challenges we ran into

One of our biggest technical challenges was connecting our Flask API to the React frontend in a secure and stable way. Chat history retention was also tricky, as we wanted users to be able to reflect on past responses but had to balance that with storage and UX considerations.

We also ran into a major setback during development. After spending hours building a key feature, we realized it wasn’t working the way we intended. It was a hard decision, but we scrapped it and rebuilt the flow from scratch. That moment tested our patience, but it ultimately led to a cleaner and more functional product.

Accomplishments that we're proud of

We’re proud that we didn’t give up after hitting a wall. Instead of forcing a broken solution, we regrouped, restarted, and came out with something that worked even better. We're also proud of how we pushed ourselves to learn new tools and build something meaningful in a short timeframe.

What we learned

We learned that it’s okay to throw out work, even if it took hours, if it helps the overall product. We also got hands-on experience with technologies that were new to some of us, like Auth0, MongoDB, and integrating AI APIs. Beyond the tech, we learned how to collaborate under pressure, keep momentum, and make tough calls as a team.

What's next for Pitch Perfect

We’d love to add support for video responses. Users could record themselves delivering their response, and the AI could analyze more nuanced aspects like facial expression, posture, and tone of voice. We think that would take the practice experience to the next level and make it even more valuable for users preparing for high-stakes conversations.