Case study — 2024 — 2026
Receptionist AI
No-code AI voice agents that call, book, and convert. A no-code AI voice-agent platform: human-like agents that handle outbound and inbound calls, qualify leads, book meetings, and trigger actions — SMS, email, WhatsApp, calendar invites — mid-call.
- Type
- saas
- Years
- 2024 — 2026
- Stack
- Node.js · Python · WebSockets · Telephony APIs · OpenAI API · ElevenLabs · PostgreSQL · Redis
The problem
Businesses miss calls, and missed calls are missed revenue. Hiring round-the-clock calling staff doesn't scale; most voice bots can't hold a real conversation, let alone complete a booking or hand off to a human.
My role
Core engineer: real-time telephony integration, automation pipelines for lead handling, and the SaaS dashboard.
The hard part
Voice AI is a latency problem before it's an AI problem. Speech-to-text, reasoning, and text-to-speech have to round-trip fast enough to feel conversational (sub-500ms call latency), while call state, lead records, and calendar writes stay consistent behind the conversation — including when callers interrupt, switch languages, or drop mid-booking. On top sit a visual flow builder, RAG knowledge bases, and Zapier/webhook actions.
Outcome
Live at tryreceptionist.ai: inbound/outbound calling in 100+ countries, 11+ languages with automatic detection, sentiment analysis, a drag-and-drop conversation flow builder, and integrations from Salesforce to Cal.com. Free and Pro plans with a 14-day trial.
