Ten thousand calls. Zero humans. One voice.
A visual studio to design, deploy, and supervise a fleet of autonomous voice agents — agents with memory, tools, and knowledge that book appointments, collect payments, and escalate to humans. Meet Ananya: Indian English with natural Hinglish, self-improving call by call via Fitty intelligence.
Not IVR. Not a bot. An actual voice workforce.
WTF Voice is a fleet of real-time conversational AI agents running on streaming speech-to-text → LLM → text-to-speech pipelines with sub-second perceived latency. It handles natural interruptions, detects when the human has finished speaking, and responds like a person — not a menu.
It handles both inbound calls (prospects, members, payments, complaints) and outbound campaigns (lead qualification, renewals, win-backs, reminders) — on local +91 numbers, in production, across 60+ WTF gym locations.
The default persona, Ananya, speaks Indian English with natural code-mixed Hinglish — because that's how members actually talk. She runs on WTF's own Hinglish speech models — in-house speech recognition and neural voice synthesis — with WTF voice-cloning for brand-specific personas.
It calls.
It closes.
It never sleeps.
Every conversation is a live streaming pipeline. The moment a member finishes speaking, the VAD fires, the transcription streams, the LLM generates, and audio begins playing back — all within 800 milliseconds of perceived silence.
Barge-in detection means Ananya yields the moment a member starts speaking mid-sentence — no robotic wait-your-turn. It feels like a person because the architecture demands it.
Sub-second response
Streaming STT → LLM → TTS with WebRTC transport. No polling, no buffering, no pauses.
Natural turn-taking
VAD-based barge-in lets members interrupt naturally. Ananya yields, listens, and re-engages.
Persona fidelity
WTF's own voice-cloning delivers a consistent, brand-calibrated voice on every call.
Everything a voice team does. Automated.
Real-time conversational engine
WTF's own streaming STT → LLM → TTS pipeline. Barge-in, turn detection, and latency-budget discipline baked in. Sub-800ms voice-to-voice, no exceptions.
Drag-and-drop agent builder
A React Flow canvas where any operator can wire conversation nodes, branch conditions, and tool calls visually. First working bot live in under two minutes — no code required.
Outbound campaign dialer
Pacing engine, answering-machine detection, automatic retries and callbacks. DND, DLT, TRAI, and consent pre-flight before any dial. Manages 10K+ calls a day without a single human in the loop.
Inbound routing on +91 DIDs
Local numbers across every WTF brand answered instantly by published agent workflows. No hold music, no IVR trees — straight to a live conversation with Ananya or any configured persona.
"Fitty" intelligence layer
Ranks the daily call list by next-best-action probability. Audits and scores every completed call automatically. Learns from outcomes to sharpen the list tomorrow.
Knowledge base, payments & escalation
RAG over pgvector for live Q&A. Razorpay payment links sent mid-call. Seamless human-agent escalation with full call context, transcript, and sentiment handed off in real time.
Agents with a goal. And a way to get better.
Every agent carries a goal and a success metric. Every call is scored deterministically — goal completion, not vibes — and the agent self-evolves with a human in the loop.
Goal + success metric, per agent
Every agent is configured with an explicit goal (book the renewal, collect the payment) and the metric that proves it. No goal, no agent.
Deterministic goal-completion scoring
The post-call eval pipeline scores every conversation on goal completion — repeatable, auditable, not a fuzzy thumbs-up.
Self-evolving — human approves the change
An in-house judge proposes a concrete improvement to the agent. A human approves it before it ships. The agent gets sharper every cycle.
Hinglish-first live-call intelligence
Telephony audio is upsampled into our in-house speech recognition before a single word is transcribed — so 8 kHz call audio is heard like studio audio.
A sacred voice-to-voice budget
Sub-second perceived latency is a hard rail, not a target. The whole pipeline is engineered to stay under it on every turn, every call.
Our voices. Zero third-party TTS bill.
An Explore catalog of in-house voices with faceted filters, inline preview, and a My Voices tab — plus clone & design new voices, in production, on our own speech stack.
Faceted voice catalog
Browse the in-house voice library and filter by language, gender, and accent. Inline preview plays each voice on the spot — pick one and ship it.
Clone & design in production
A My Voices tab for the voices you own. Clone a brand voice or design a fresh persona on our in-house speech stack — live, not a sandbox.
Owned voices, owned cost
Because every voice runs on our own neural synthesis, there's no per-character third-party TTS invoice. Scale voices without scaling a vendor bill.
Brain · Voice · Ears, tuned live
Compose an agent from an LLM brain, an in-house TTS voice, and in-house STT ears — and watch the cost and latency chip recompute as you tune.
Our voice engine, in your stack.
A metered Voice Engine API for developers. Key-auth, per-character metering, idempotent wallet billing, and per-key rate limits — wired the way you'd build it yourself.
Scoped API-key auth
Issue scoped keys per app or environment. Each key authenticates the request and carries only the permissions you grant it.
Per-character metering
Usage is metered per character synthesized — transparent, predictable, and visible in real time on every key.
Idempotent wallet billing
Charges hit a prepaid wallet idempotently — retried requests never double-bill. Top up once; spend safely.
Per-key rate limits
Every key gets its own rate limit, so one noisy integration can never starve the rest of your fleet.
Every touchpoint in the member lifecycle. Handled.
WTF Voice doesn't replace a telecaller for one use case — it replaces the entire inbound and outbound calling function across every stage of the member journey.
Lead qualification & sales
Dials inbound enquiries within seconds, qualifies interest, pitches the right plan, and converts to a trial or paid membership.
Renewals & payment collection
Proactive renewal calls before expiry, overdue payment collection with live Razorpay link delivery, and confirmation follow-ups.
Reminders & check-ins
Visit reminders, BMI & re-test check-in calls, class schedule confirmations, and personalised health nudges based on member data.
Win-back & birthday calls
Churned member win-back campaigns with personalised offers, birthday call sequences, and lapsed-visit re-engagement flows.
The voice-agent platform built to outrun the entire category.
No per-minute markup, no black box, no lock-in. Proprietary and owned end-to-end, and compliance hard-railed in code — not a settings panel.
Owned end-to-end
Every layer — agents, models, voice, data, infrastructure — is proprietary and built in-house. No black boxes we don't control, no per-seat tax.
India-first, Hinglish-native
In-house Indic speech models, local +91 numbers, code-mixed scripts — purpose-built in-house for a billion-member market, not retrofitted.
Compliance hard-railed
TRAI TCCCPR 2018 + Feb 2025 amendment, DLT registration, DND scrubbing, and DPDP 2023 consent gates enforced in code before any dial fires.
Model-agnostic
Swap any speech or language model in one config line — our own model router adapts as the leaderboard shifts, with zero rewrites.
Multi-provider telephony
Twilio, Vonage, Plivo, IVR Solutions — all supported. Bring your own numbers, your own SIP trunk, your own redundancy strategy.
Production, not a demo
Versioned, CI-deployed, runbook-backed. 60+ gyms live, 10K+ calls a day, shipping weekly. This is not a proof of concept.
Every component chosen for latency, cost-efficiency, and India-scale. Model-agnostic at every layer — swap without touching the pipeline.
Voice is one part of a compounding loop.
Every call surfaces intent. That intent feeds messaging, which feeds video, which drives the leads that fill tomorrow's call list. One loop. No leakage.
WhatsApp CRM
Call outcomes, callbacks, and consent sync straight into the inbox. A self-hosted Meta Cloud API CRM — no BSP, no markup.
Video Engine
What converts on the call becomes the brief for the next batch of Reels. 100 brand videos a week, for 30 cents each.
Put Ananya
to work today.
Deploy the autonomous voice workforce across your locations — or deploy it in your own private cloud.
PROPRIETARY · PRIVATE-CLOUD · PRODUCTION-GRADE