PitchBox

Inspiration

What it does

Pitchbox streamlines automated product demo capture by recording headless browser sessions and generating voiceovers for fast, engaging presentations. This addresses the challenge of sharing real interactive product experiences, making demo workflows more efficient and accessible for teams and hackathons. Technical architecture: Pitchbox consists of a Next.js/Node.js monorepo. The backend spins up Chromium via Puppeteer and encodes video with ffmpeg, then adds AI-generated voiceovers using the ElevenLabs API. Tigris is integrated as a transactional cloud database and storage layer, managing metadata and storing uploaded demo assets for scalable, secure retrieval. Sponsor tools/protocols: Daytona Cloud: For isolated web app sandboxes and automated deployment via Daytona SDK.

ffmpeg/Xvfb/Puppeteer: For browser automation, video capture, rendering.

ElevenLabs: For realistic AI voiceover generation. Integrated in backend scripts to automatically synthesize spoken pitch audio from scripts or demo content.

Tigris: For asset storage, user/session metadata, and fast querying. Linked via the backend for seamless asset management and retrieval.

Each integration is orchestrated through API logic, with sponsor tools leveraged for their specialization in environments (Daytona), voice synthesis (ElevenLabs), and data storage (Tigris), allowing end-to-end demo automation.

Built With

• anthropic
• daytona
• eleven-labs
• tigris