{"slug":"oriserve-s-generative-ai-voice-bots-cut-cost-to-serve-50-and-lift-a-telecom-client-s-customer-retention-75","url":"https://findausecase.com/use-cases/oriserve-s-generative-ai-voice-bots-cut-cost-to-serve-50-and-lift-a-telecom-client-s-customer-retention-75","title":"Oriserve's generative AI voice bots cut cost to serve 50% and lift a telecom client's customer retention 75%","description":"Oriserve, an India-based generative AI customer engagement startup, built multilingual AI agents and conversation analytics on Amazon Bedrock supporting more than 120 languages and dialects. Using large language models including Claude 3.5 and Llama 3-70B, Oriserve reduced its cost to serve by 50% and cut LLM fine-tuning time from weeks to days, while its voice bot achieved 90% automatic speech recognition accuracy with 60% fewer word errors than industry standards. Indian telecom provider Vi used Oriserve's generative AI agents to immediately engage customers during the narrow window after a number-port request is created, increasing customer retention rates by 75%; other customers included an education technology company that tripled qualified leads and a global retail brand that saw a 10% revenue boost.","company":"Oriserve","industry":"Technology & Software","country":"India","aiCapabilities":["Conversational AI","Large Language Models","Speech & Audio AI","Generative AI"],"technology":["Amazon Bedrock","Amazon Titan","Amazon EC2 Auto Scaling","Amazon Elastic Kubernetes Service"],"deployment":"Unknown","problemStatement":"As a bootstrapped startup, Oriserve needed to avoid the higher costs of on-premises infrastructure while ensuring compliance across closely regulated industries like BFSI, real estate, telecommunications, education, and travel, and needed a cloud provider that could optimize generative AI application development and deliver the scalability and reliability required for rapid go-to-market.","solutionApproach":"Oriserve used Amazon Bedrock to build text and email bots on large language models such as Anthropic Claude 3.5, Meta Llama 3-70B, and Amazon Titan Text Embeddings, deployed as microservices on Amazon EC2 Auto Scaling instances managed with Amazon Elastic Kubernetes Service, and developed custom voice models with automatic speech recognition.","businessValue":"Oriserve reduced its cost to serve by 50% and cut LLM fine-tuning time from weeks to days using Amazon Bedrock's managed services; its voice bot achieved 90% accuracy across dialects with 60% fewer word errors than the industry average, and its generative AI bots for telecom client Vi increased customer retention rates by 75%, contributing to Oriserve's 50% annual revenue growth.","evidence":{"band":"high"},"sourceUrl":"https://aws.amazon.com/solutions/case-studies/oriserve-case-study/","dates":{"publishedAt":"2026-09-15T09:05:30.513Z","publishedAtSource":"pipeline","updatedAt":"2026-09-15T09:05:30.513Z"},"license":"Open for reading and citing with a link to https://findausecase.com/use-cases/oriserve-s-generative-ai-voice-bots-cut-cost-to-serve-50-and-lift-a-telecom-client-s-customer-retention-75. Bulk republication requires permission."}