DEV Community

VoiceFleet
VoiceFleet

Posted on • Originally published at voicefleet.ai

Building AI Receptionist Integrations with Dentally and Flipdish APIs

Most AI receptionist products stop at "answer the call and email you a transcript." That's not automation — that's delegation to a slower process.

We've been building integrations between our voice AI and the actual business tools Irish companies use daily: Dentally (dental practice management) and Flipdish (restaurant ordering platform).

The Technical Challenge

The interesting engineering problem isn't the voice AI itself — it's the real-time decision loop during a live call:

  1. Caller says "I need to book an appointment for Thursday"
  2. AI queries Dentally API for Thursday availability (< 500ms budget)
  3. AI presents options conversationally
  4. Caller confirms
  5. AI creates booking via API + sends SMS confirmation

All of this needs to happen within natural conversation flow. You can't have a 3-second pause while your API call resolves.

Flipdish Integration Pattern

Restaurant ordering has a different challenge: menu complexity. A restaurant might have 200+ items with modifiers, sizes, and combos. The AI needs to:

  • Map spoken items to exact menu entries ("large pepperoni with extra cheese")
  • Handle modifications ("no onions, add mushrooms")
  • Calculate pricing in real-time
  • Push the complete order to the kitchen display

We sync the full Flipdish menu on a schedule and keep a local cache for sub-100ms lookups during calls.

Key Takeaway

If you're building voice AI for business use cases, the voice model is maybe 30% of the work. The other 70% is integrations, edge cases, and making API calls fast enough that the conversation feels natural.

Would love to hear from others building real-time API integrations into conversational AI — what latency budgets are you working with?

Top comments (0)