Stagee Core · Voice and language infrastructure

The technology that listens, understands, and acts.

Stagee Core is the voice and language engine that powers the agent. Built specifically for hospitality, from speech recognition to confirming the reservation in the right system.

< 80ms

ASR recognition

2+

Native languages

24kHz

Sample rate

2

Native integrations

Not a chatbot

Not a chatbot. A reasoning agent.

Most automated answering solutions follow scripts. Stagee Core does not. It reads the caller real intent, adapts the response to the conversation context, and acts on the reservation system in real time.

It is built on a low-latency voice architecture combining speech recognition (ASR), natural language understanding (NLU), and speech synthesis (TTS), all optimized for the rhythm of a real phone conversation.

  • ASR · Speech recognition

    ASR · sub-80ms

  • NLU · Semantic understanding

    Semantic NLU

  • Action · Booking in the system

    TheFork and UMAI API

  • TTS · Voice synthesis

    Neural TTS · pt-PT

Technical capabilities

What Stagee Core can do.

Hears like a person hears.

Speech recognition with sub-80ms latency. Handles regional Portuguese accents, dining room background noise, and hospitality-specific terms (dishes, ingredients, allergens).

Understands what the client actually means.

Semantic NLU with restaurant-specific prompt engineering. Does not follow scripts.

Knows when to hand off the call.

When the conversation exceeds what the agent can handle, it transfers to the team. The written call summary is available in the portal.

Follows the conversation from start to finish.

Persistent context throughout the call. Automatic post-call summary with intent and collected data, accessible in the portal.

A voice that sounds like a person.

Neural voice synthesis in European Portuguese. Natural intonation, no artificial accent.

Tech stack

Built on a best-in-class stack.

ASR · speech recognition

Speech recognition with sub-80ms latency for PT + EN.

< 80ms · PT + EN native

NLU · semantic understanding

Semantic reasoning over caller intent, in native pt-PT.

Semantic NLU

TTS · speech synthesis

Neural speech synthesis in European Portuguese. Natural voice with prosody adapted to hospitality.

Neural voice · pt-PT

Real-time voice orchestration

Infrastructure that coordinates ASR, NLU and TTS with sub-second latency.

Real-time · multi-tenant

Telephony and SMS

Phone calls and confirmation SMS. Supports Portuguese numbers and routing to the team.

SIP · SMS · PT numbers

Database

Storage for configuration, call history and metrics. EU hosted.

EU hosted · GDPR

TheFork and UMAI integration

Native TheFork and UMAI integration.

Reservations made by the agent land directly in TheFork or UMAI. No manual work, no risk of error.

  • Reservation registered directly in the restaurant reservation system, TheFork or UMAI, without human intervention
  • Real-time availability check before confirming
  • Confirmation SMS sent automatically to the caller
  • Stagee portal updated with each reservation in real time
  • Full compliance with the restaurant reservation policies

Call flow

  1. 1Client calls
  2. 2Intent detected
  3. 3Data collected
  4. 4Availability verified
  5. 5Reservation confirmed

Native TheFork and UMAI integration

Grottu · production

Numbers running in production.

Real technical metrics from the agent live at Grottu, Porto.

2s

End-to-end latency

< 1%

Human handover rate

99.9%

Uptime

2+

Months in production

Stagee Core · roadmap

What is shipping, and what is next.

In production
  • Voice
  • Reservations
  • Bilingual PT + EN
  • Events · group qualification
In development
  • Operations
  • Commercial
  • Sentiment analysis
  • Guest memory
  • Multichannel
  • Website chat
  • Warm transfer
Roadmap
  • Stagee Care · healthcare
  • Stagee Sports · gyms and sports clubs

The technology is ready. So is the agent.