Ringg, a voice and chat agent platform serving large consumer businesses in India, built an enterprise agent platform around OpenAI models to handle high call volumes without simply adding human agents. The company reports that its agents now handle more than 7 million connected calls per month, with an average customer satisfaction score of 4.8, and resolve up to 65% of routine customer inquiries without human involvement.

The platform spans voice, chat, WhatsApp, and the web. For real-time interactions, Ringg uses GPT-5.6 Luna and other models to interpret customer requests, select tools, and guide customers through multi-step workflows. An orchestration layer executes actions across CRMs, ticketing platforms, payment systems, scheduling tools, and internal APIs, escalating cases to a human with a conversation summary when needed. A knowledge system combines structured filtering with semantic retrieval across datasets, PDFs, CSVs, and business documents. Ringg can also divide work among specialized subagents for qualification, support, verification, scheduling, and escalation, while maintaining a consistent conversation across channels.

Ringg routes work to a specific OpenAI model based on task needs. GPT-4.1 handles most real-time voice and chat traffic. GPT-5.6 Luna is used when its performance, latency, or price-performance profile better suits a request. GPT-5.6 Terra handles post-call analysis, including summaries and sentiment classification. GPT-5.6 Sol supports evaluation, prompt improvement, and model-as-judge workflows. For longer interactions, the system creates a structured summary when context approaches approximately 80,000 tokens, preserving important information without repeatedly sending the entire history.

Ringg tests models using historical conversations and simulated customer flows before production deployment. In one evaluation, GPT-5.6 Terra outperformed Gemini 2.5 Flash for post-call analysis, so Ringg moved summaries and sentiment classification to Terra. Terra also outperformed Gemini 2.5 Flash on regional language accuracy, with up to 97% accuracy on common regional languages. Models that pass offline testing are introduced to a small share of production traffic before rollout expands. In production, the router monitors latency and endpoint health across regions and shifts traffic when an endpoint becomes unavailable or crosses a latency threshold. Migrating certain real-time workloads from GPT-4.1 to GPT-5.6 Luna reduced model costs by approximately 90%.

Customer results cited include Policybazaar, one of India’s largest online insurance platforms, which connects more than 57,000 customer requests through Ringg, with 67% of calls handled without human intervention and average response time falling from 8-12 minutes to under 60 seconds. Practo, a global healthcare platform, achieved an 85% first-call resolution rate, response times below three seconds, and a 70% decline in operating costs, with more than 1,000 appointment bookings completed each day. Groww, an online investment platform, resolves 72% of inbound queries related to IPOs, futures, and options entirely through self-service, with an average handling time of two minutes.

Ringg is also developing browser agents using OpenAI‘s computer-use capabilities for platform onboarding, Know Your Customer (KYC) processes, IT troubleshooting, on-call incident support, and claims processing. The company is building a context layer to preserve information across channels and interactions, so a customer can begin a request over voice, continue on WhatsApp, and finish in a browser without repeating details.