Ringg Leverages OpenAI’s GPT‑5.6 to Automate 65% of Customer Calls, Cutting Costs by 90%

What happened, who, and when

On September 23, 2026, Ringg announced that its AI‑driven contact‑center platform, powered by OpenAI’s latest GPT‑5.6 models, is able to settle up to 65 % of routine customer calls without human intervention. The startup, based in the Asia‑Pacific and Oceania region, claims the new architecture delivers multilingual support across voice, chat, WhatsApp and web at a fraction of the cost of earlier GPT‑4.1 deployments.

Concrete details of the deployment

• Scale and performance: The system processes over 7 million connected calls each month, achieving an average customer satisfaction (CSAT) rating of 4.8.
• Cost efficiency: By shifting suitable real‑time workloads from GPT‑4.1 to GPT‑5.6, Ringg reports a near‑90 % reduction in model expenses while preserving required latency and quality.
• Model orchestration: Ringg’s platform routes each interaction to the most appropriate OpenAI model: GPT‑4.1 handles the bulk of voice and chat traffic; GPT‑5.6 “Luna” is used when its speed, price‑performance or capabilities are a better fit; GPT‑5.6 “Terra” powers post‑call analysis such as summarization and sentiment classification; GPT‑5.6 “Sol” supports evaluation and prompt‑tuning workflows.
• Tool integration: The agents invoke a suite of enterprise tools—including CRMs, ticketing systems, payment gateways, scheduling apps and internal APIs—to complete multi‑step tasks like policy checks, account lookups, appointment bookings, and handoffs to specialists with full context.
• Knowledge handling: A hybrid knowledge system blends structured filtering with semantic retrieval across PDFs, CSVs and other business documents, enabling agents to answer queries with up‑to‑date enterprise information.
• Evaluation pipeline: Before production rollout, Ringg runs historical conversation data and simulated flows through an offline evaluation platform. Results feed back into prompt‑improvement cycles, creating a continuous quality‑cost loop. In one test, GPT‑5.6 “Terra” outperformed Google’s Gemini 2.5 Flash on summary accuracy and sentiment analysis, achieving up to 97 % accuracy on regional language inputs.
• Production safeguards: A routing layer monitors latency and endpoint health across regions, automatically shifting traffic away from degraded nodes. Versioned deployments and specialized alerts isolate issues, protecting both performance and economics.
• Customer outcomes:
• Policybazaar: Handles more than 57 000 daily requests; 67 % of calls are resolved without a human, and average response time dropped from 8‑12 minutes to under 60 seconds (≈ 88 % improvement).
• Practo: Achieved an 85 % first‑call resolution rate with sub‑3‑second response times, cutting operating costs by 70 % and processing over 1 000 appointment bookings per day.
• Groww: Resolves 72 % of inbound queries about IPOs, futures and options via self‑service, with an average handling time of two minutes.
• Browser agents: Leveraging OpenAI’s computer‑use capabilities, Ringg is prototyping agents that can guide users through KYC, IT troubleshooting, incident support and claims processing directly in a web browser, preserving context across voice, messaging and web channels.
• Founder’s remarks: “Model quality is only part of the equation. We also need low latency, reliable tool use, strong instruction following, and economics that work at scale. OpenAI gave us the balance we needed,” said Siddharth Tripathi, Ringg co‑founder. He added that OpenAI’s dedicated Slack support and direct engineering access have accelerated migration and reduced engineering uncertainty.

Industry context and competitive landscape

Ringg’s results illustrate a broader shift in contact‑center economics: enterprises are replacing headcount‑driven scaling with AI agents that can handle high‑volume, multilingual interactions at dramatically lower cost. The company’s internal benchmarking found OpenAI’s GPT‑5.6 suite to outperform alternatives such as Google’s Gemini 2.5 Flash on instruction adherence, tool‑calling reliability and multilingual accuracy—key criteria for large‑scale customer‑service deployments. OpenAI’s recent focus on startup enablement, including streamlined API access and responsive engineering support, appears to be paying off as more regional players adopt its models to meet rising demand for real‑time, outcome‑driven automation.