Back to demo hub

AI Voice demo

University admissions call

Product simulation — AI-generated demonstration voice. The scripted playback uses fictional data with neural text-to-speech (ElevenLabs for Urdu, OpenAI for English and Arabic). The live two-way section below is a real AI conversation on fictional records; neither is a real customer call.
Ready to play
Play the Voice story
System

Incoming call from +971 55 123 4567

Noise
0:001:20

Voice provenance & mix

This is a product simulation in English (en-US). Two distinct synthetic voices are used so the AI agent and the customer are never confused with each other.

product simulation
Quantara AI Agent
female
Provider: Quantara AI Voice (neural TTS)
Model: openai/gpt-4o-mini-tts
Voice: Shimmer
Customer
male
Provider: Quantara AI Voice (neural TTS)
Model: openai/gpt-4o-mini-tts
Voice: Ash
Disclosure

Ambience: Subtle indoor room tone on the customer track only. Speech is generated at playback time by the same speech stack the product uses for live calls; the conversation script itself is fictional demo content.

Post-call timeline

Transcript

Full speech-to-text with speaker diarisation.

Summary

Resolved query, sent reference reminder, no handoff.

Wrap-up

Reason: reference_reminder_sent.

QA

Tone 5, accuracy 5, SLA 5, empathy 4.

SLA

First response 2s, resolution pending document.

Insight

Top topic: missing reference letter.

Full interaction

Click any step to inspect the AI decision and operational context.

AI Voice IVR for English and Arabic support operations

The demo above is a scripted walkthrough of a real call path. Below is how AI Voice works in production, what is available today and what runs as a guided pilot.

What is an AI Voice IVR?

An AI Voice IVR answers an inbound call in natural language instead of a keypad menu. Quantara Flow AI greets the caller, identifies who they are, understands the reason for the call, answers from your own knowledge base, performs approved actions and transfers to a human when a person is needed — then writes the call back into the case, Customer 360 and reporting.

  • No “press 1 for billing” trees — the caller says what they need
  • Intent, sentiment and language detected on the call
  • Every eligible call creates or links to exactly one case

Arabic AI Voice Agent

The Arabic AI Voice Agent handles inbound calls in Arabic, understands mixed Arabic-English speech and answers from your Arabic knowledge articles. Availability follows the configured Voice Profile: Gulf/Saudi Arabic and Modern Standard Arabic profiles are available, and dialect steering is enabled per workspace after a call test.

English and Arabic AI Voice on one number

One Voice agent can serve both English and Arabic callers. The language is detected at the start of the call, the caller can switch mid-conversation, and the transcript, wrap-up reason and QA score are recorded in the language of the call so reporting stays comparable across languages.

Saudi Arabic Voice configuration (guided pilot)

Saudi Arabic Voice configuration runs as a guided pilot. We configure the Voice Profile, pronunciation dictionary and dialect steering with your team, then validate it on real calls with a fluent speaker before it goes live on a public number. We do not present dialect coverage as production-proven until that call acceptance is complete.

Emirati Arabic Voice configuration (guided pilot)

Emirati Arabic Voice configuration is also delivered as a guided pilot. Emirati vocabulary, names and place pronunciations are tuned per workspace and validated on real calls with a fluent speaker. Until that acceptance evidence exists the configuration remains pilot-only rather than a published capability.

AI call center for UAE and Saudi Arabia support teams

Support teams in the UAE and Saudi Arabia typically run bilingual queues, long peak hours and mixed WhatsApp and phone contact. Quantara Flow AI answers the routine calls, keeps the customer context shared with digital channels, and gives supervisors interval reporting, service-level tracking and capacity views for the phone queue.

Intelligent IVR built in Voice IVR Studio

Voice IVR Studio is a visual builder for the call flow: greeting, identity verification, business hours, language selection, knowledge answers, approved tool calls, queue routing and transfer. Only node types with a tested runtime executor can be published, so a flow cannot go live with a step the Voice runtime cannot actually execute.

SIP integration with your existing carrier

Quantara Flow AI connects over SIP, so you can keep your existing numbers and carrier. Inbound calls are routed to the AI agent, and transfers go back out to your queues. SIP onboarding is currently delivered as an assisted onboarding (beta) with our team, not self-service.

PBX integration and human transfer

The AI agent can sit in front of an existing PBX or contact-centre platform. When the caller asks for a person, the confidence is too low, verification fails or a promised action needs approval, the call transfers to the configured queue or department with the transcript, identified customer and case already attached — so the agent does not restart the conversation.

Voice QA on every call, not a sample

Instead of reviewing a small sample, the platform scores calls against your published QA rubric for the channel, language and AI-or-human mode. Supervisors see the score, the evidence behind it and the calls worth listening to. Post-call QA on the durable job path is in beta while its recorded acceptance evidence is completed.

Automatic call summaries and wrap-up

After the call ends, a durable post-call job finalises the transcript, writes the summary, selects one active wrap-up reason from your own configured list, applies case status and SLA, and updates Customer 360 and reporting. The AI can only pick a wrap-up reason you have configured — it never invents a new code, department or escalation path.

RAG-grounded Voice answers

Voice answers are retrieved from your own knowledge articles and SOPs with hybrid keyword and vector search, then generated with citations. If nothing relevant is retrieved or confidence is insufficient, the agent does not guess — it escalates or transfers to a human and records the knowledge gap for your team to fill.

Which languages and dialects can AI Voice speak?

  • EnglishAvailable — Browser and SIP sessions in production
  • Arabic — Saudi · العربية — سعوديAvailable — Dialect steering enabled per workspace after a call test
  • Arabic — Gulf · العربية — الخليجيةAvailable — Dialect steering enabled per workspace after a call test
  • Arabic — Modern Standard Arabic · العربية الفصحىAvailable
  • Arabic — Emirati · العربية — إماراتيAvailable — Dialect steering enabled per workspace after a call test
  • Urdu · اردوAvailable — Runs as an English/Urdu pair with fluent-speaker validation per workspace
  • Spanish · EspañolAvailable
  • French · FrançaisAvailable

Voice language and dialect availability depends on the configured Voice Profile and provider. Dialect coverage is enabled per workspace after a call test.