Pitchy AI

Pitchy AI lets users submit audio or text pitches and get AI analysis, metrics, and improved rewrites, plus users hear both original and enhanced versions using Google Gemini and ElevenLabs.

Comment

Inspiration

Pitchy AI was inspired by learning about the capabilities of ElevenLabs API and their unique use cases for emotive voice overs which we want to utilize in teaching and improving the structure of pitches for oncoming entrepreneurs. Through research we also found that AI agents used in business and marketing are only done via LLMs even when we know there are so much more nuances of voices and how a person projects themselves that highlight the quality of their pitch.

What it does

Pitchy AI is a web-based application that enables users to submit pitches, either as live audio recordings or typed text, and receive AI-driven analysis, performance metrics, and an optional improved rewrite. The platform integrates Google Gemini for language understanding and pitch evaluation, and ElevenLabs for high-quality AI voice synthesis, allowing users to hear their pitch back in both its original and improved forms.

How we built it

** Tech Stack ** Frontend HTML CSS Tailwind CSS Vanilla JavaScript

Browser APIs MediaRecorder getUserMedia fetch localStorage

Backend Node.js Express Multer CORS dotenv

AI / APIs ElevenLabs (Speech-to-Text) Google Gemini (AI Feedback)

Challenges we ran into

One of our biggest issues was definitely running and testing the two AIs and making sure they were communicating with each other properly. Not only did we run our of tokens for some models, but Gemma4 was also only usable by one programmer due to the software requiring a dedicated GPU just for local processing.

Accomplishments that we're proud of

One of out biggest accomplishments was actually getting ElevenLabs to talk from the prompt that Gemini got from our recording. Connecting these two is something no one has probably never done before and with this possibility, the app can scale up in functionality.

What we learned

API Calls via Gemini, Gemma4, and ElevenLabs, Frontend with HTML and JS, and Backend with node servers

What's next for Pitchy AI

Next for Pitchy AI is implementing different emotive voices via ElevenLabs so the user can try and test how their pitch with a different voice. We also want to implement Gemini or Gemma4 to implement some of the suggested changes to the pitch and in turn ElevenLabs voicing over this new and improved pitch to allow the user to test multiple different approaches as they wish.

Built With

Share this project:

Updates