Ringg Automates 65% of Customer Inquiries with AI Agent Based on OpenAI GPT-5.6

India-based voice and chat agent platform Ringg has adopted OpenAI's GPT-5.6 model to significantly improve customer service operational efficiency. Compared to the previous GPT-4.1, it reduced model costs for specific real-time workloads by approximately 90% while maintaining quality and latency.

Multi-Channel Agent Architecture

Ringg utilizes models such as GPT-5.6 Luna across various channels including voice, chat, WhatsApp, and web to interpret customer requests and select tools. The orchestration layer integrates with CRM, payment systems, and booking tools to execute tasks, escalating to human agents with conversation summaries when necessary. Additionally, GPT-5.6 Terra is used for post-call summaries and sentiment analysis, while GPT-5.6 Sol is used for evaluation and prompt improvement.

Performance Validation and Cost Reduction

Ringg evaluates models using historical conversation data and simulations before deploying to production. Notably, GPT-5.6 Terra demonstrated superior performance (up to 97% accuracy) compared to Gemini 2.5 Flash in mixed-language conversations. Through these optimizations, it handles over 7 million connected calls per month, achieving an average Customer Satisfaction Score (CSAT) of 4.8.

Key Customer Achievements

• Policybazaar: Handles 67% of insurance inquiries without human intervention, reducing response time from 8–12 minutes to under 60 seconds.
• Practo: Achieved a 85% first-contact resolution rate for medical appointments, reduced operational costs by 70%, and completes over 1,000 appointments daily.
• Groww: Resolves 72% of investment-related inquiries via self-service, with an average handling time of 2 minutes.

Ringg plans to develop browser-based agents using OpenAI's computer-use capabilities and build a layer that maintains context across channels in the future.