VocalPoint

inspiration • wanted to make real-time ai call insights easy and accessible

what it does

• transcribes live or uploaded audio
• analyzes sentiment and persona in real time
• surfaces actionable insights and compliance alerts on a dashboard

how we built it

• frontend: react + material ui, hosted on aws amplify
• backend: fastapi, dockerized and deployed with aws app runner
• genai features: amazon transcribe and amazon comprehend for speech-to-text, bedrock (claude) for nlp tasks
• stateless, real-time data flow via SSE

challenges we ran into

• integrating real-time audio streaming with web ui
• connecting frontend and backend for live updates
• prompt engineering to get concise, actionable outputs from claude

accomplishments that we’re proud of

• full aws-native, scalable deployment in a short time
• seamless real-time transcription and sentiment analysis
• clean, modern dashboard with live genai features
• minimal setup, easy to demo and extend

what we learned

• aws services are fast to set up and integrate
• bedrock (claude) is flexible for various nlp tasks

what’s next for vocalpoint

• add persistent storage for analytics and history
• expand persona modeling and insights features
• support more languages and advanced compliance guardrails

Built With

• amazon-web-services
• amplify
• claude
• react
• s3
• sse