Gupshup Releases Self-Serve Voice AI Platform with Custom Model Selection and Multichannel Support

Conversational software provider Gupshup has launched a self-serve Voice AI Platform designed to automate inbound and outbound telephony workflows for businesses. The service connects existing text-based automations on WhatsApp, RCS, and SMS to live voice agents, while letting teams choose their own speech-to-text, text-to-speech, and language models. The platform is available globally with pricing beginning at $0.035 per minute alongside 100 promotional minutes.

Bridging WhatsApp, RCS, and SMS Automations into Live Voice

Many enterprise support systems run automated assistants on text channels such as Short Message Service (SMS), Rich Communication Services (RCS), and WhatsApp. These setups traditionally operate separately from call centers, requiring distinct logic, routing rules, and backend integrations for phone conversations.

The new platform links those existing text paths directly into voice systems. When a customer interaction shifts from a messaging app to a telephone call, the platform can share the underlying workflow rules across both mediums.

This unified setup prevents businesses from having to construct parallel conversational flows from scratch. A customer who begins an interaction over text can trigger or continue that process through an automated phone call without the company rebuilding the application logic.

Modular Architecture Across Speech Recognition, Reasoning, and Voice Synthesis

Real-time voice agents operate by linking three distinct computational layers together in sequence:

• Speech-to-Text (STT): An acoustic model transcribes the user’s spoken audio into text in real time.
• Language Models (LLM): An artificial intelligence model processes the text query, consults business data or prompt rules, and generates a written response.
• Text-to-Speech (TTS): A voice synthesis engine converts the generated text back into an audio stream that plays over the phone line.

Gupshup’s platform allows businesses to select custom models for each of these three steps rather than locking developers into a single default provider.

Choosing specialized components helps teams balance latency, cost, and linguistic requirements. For example, a business can route audio through an STT engine trained for specific regional accents, while sending the text to a specialized language model fine-tuned on internal company documentation.

Inbound Call Handling and Outbound Dialing Mechanics

The system supports two primary calling patterns: inbound response and outbound automation.

Inbound workflows handle incoming phone calls to replace or augment traditional interactive voice response menus. Instead of forcing callers to press numbers on a keypad to navigate nested options, the conversational voice agent interprets spoken requests directly and answers questions or executes account actions.

Outbound workflows initiate automated phone calls to customers based on system triggers. Companies use outbound dialing for time-sensitive notifications, such as appointment confirmations, delivery tracking updates, payment reminders, and account verification steps.

Self-Serve Deployment and Per-Minute Pricing Structure

The platform uses a self-serve onboarding model. Rather than requiring multi-week procurement cycles or custom engineering contracts, teams can configure telephony numbers, connect model keys, and launch agents directly through the interface.

Gupshup set the baseline usage price at $0.035 per minute. The company includes 100 promotional minutes for testing and initial deployments, with the service open to developers worldwide.