Ringg AI is a conversational-AI platform for businesses that runs AI agents over voice calls, chat, WhatsApp, and web. You configure an assistant in the dashboard and trigger it through a REST API for inbound calls, outbound calls, web calls, and bulk campaigns, with post-call analytics and transcripts retrievable afterwards.
For builders assembling their own voice pipeline, Ringg also exposes its Parrot speech-to-text model as a standalone streaming API. Parrot targets real-time voice agents and handles English, Hindi, and Hindi-English code-mixed speech, with a typical streaming latency of 60 ms. A Python SDK (the ringglabs package) wraps the API and works with Pipecat using built-in voice activity detection events. The model is proprietary and served only through the API; production access is granted on approval.
Features
- Programmable voice agents: start single or scheduled outbound calls, answer inbound calls, and run web calls against an assistant configured in the dashboard, passing dynamic prompt variables per call
- Campaign API: upload a CSV contact list and run bulk calling campaigns with configurable calling windows and retry behaviour
- Call history and analytics: retrieve transcripts, call status, duration, recordings, and performance metrics through the API
- Webhooks: receive events for call started, call completed, recording available, analysis, and processing completion
- Parrot streaming STT: low-latency streaming speech recognition for English, Hindi, and code-mixed speech, reported at 60 ms latency with a 7.27% normalised word error rate
- Audio input: accepts WAV, MP3, FLAC, M4A, OGG, and OPUS, with 16 kHz or higher sample rates recommended
- Concurrency: up to 30 concurrent streaming connections included as standard
- Pipecat integration: the Python SDK plugs into Pipecat pipelines using its VAD events
- Web widget: embeds a voice or chat agent in a website without custom front-end code
