AI Receptionist Platform

NOV 2025 – FEB 2026


Summary

  • Multi-tenant SaaS architecture with full data isolation per organization
  • Text chat widget embeddable on any website plus Twilio-powered inbound voice calls with Deepgram STT and ElevenLabs TTS
  • RAG-powered knowledge retrieval using Upstash Vector for semantically grounded, company-specific responses
  • Customizable AI personas with configurable system prompts, greeting messages, tool access, and brand voice
  • Calendar & webhook integrations for appointment booking (Google/Outlook) and real-time event-driven workflows
  • Full admin dashboard with role-based access, chat log analytics, knowledge base management, and a super-admin tier
  • Modern tech stack — Next.js 16, React 19, TypeScript, Prisma, Tailwind CSS, Vercel AI SDK, Vite (widget), Zod validation

AI Receptionist is a full-stack, multi-tenant SaaS platform that provides businesses with an AI-powered virtual receptionist capable of handling customer interactions around the clock via both text chat and voice.


Built on Next.js 16 with React 19, TypeScript, and Tailwind CSS, the platform allows organizations to deploy a customizable AI assistant on their website through an embeddable chat widget or connect it to inbound phone calls via Twilio.
Each tenant have their own space for their scoped personas, knowledge bases, vector embeddings, tools, and settings. It is suitable for white-label deployment across diverse business verticals.

The system leverages Retrieval-Augmented Generation (RAG) with Upstash Vector to ground AI responses in each organization’s proprietary knowledge base, and integrates with the Vercel AI SDK and Groq for LLM inference.

Voice interactions are powered by Deepgram for speech-to-text and ElevenLabs for text-to-speech; with integration with Twilio, clients are able to have natural phone conversations with the personas.

Additional capabilities include calendar booking (Google Calendar/Outlook), webhook-driven integrations, configurable pre-chat forms, automated chat summarization, and a Model Context Protocol (MCP) layer for extensible tool use. The admin dashboard also provides company-scoped management of personas, knowledge chunks, chat logs, callbacks, and settings.

More PROJECTS