Inspiration
Small businesses lose customers when they miss phone calls, but setting up a voice agent usually requires filling out long forms with business information. We wanted to make onboarding effortless by letting AI learn directly from a company's website and existing documents, then asking only about information it couldn't find.
What it does
Frontdesk automatically builds the knowledge a voice agent needs before answering customer calls. Business owners simply upload their website and documents, then have a short conversation with an AI onboarding agent. Frontdesk extracts business information, verifies it with the owner, and organizes it into a structured knowledge base with editable source citations, allowing voice agents to answer customer questions accurately.
How we built it
We built Frontdesk using Next.js, React, Tailwind CSS, Supabase, and Postgres, with Claude powering the onboarding agent. The system combines website scraping, document parsing, and conversational AI to extract business information. A deterministic state machine manages the onboarding flow while every extracted fact is linked back to its original source, ensuring the final knowledge base is both accurate and easy to edit.
Built With
- anthropic
- claude
- google-calendar-api
- javascript
- next.js
- node.js
- oauth2
- pdf.js
- postgresql
- react
- supabase
- tailwind-css
- vitest

Log in or sign up for Devpost to join the conversation.