Asia’s Leading regional Bank Enhances Customer Service with Tetherfi’s Edge AI Solution
Asia’s Leading regional Bank, renowned for its innovation and customer-first approach, is a top financial institution committed to redefining the banking experience through digital transformation and AI-driven service excellence.
Challenges/Opportunities
As part of their GenAI innovation roadmap, the bank launched an
initiative aimed at elevating customer service experiences by
empowering their Contact center agents with real time STT
transcription, emotion detection, intelligent interaction
assistance, and post-call summarization. A critical component for
success was the need for accurate, low-latency transcription of live
conversations between agents and customers.
The client’s environment required a solution that could perform
efficiently within existing infrastructure constraints. With limited GPU
resources, the focus shifted to deploying a transcription engine that
was fully optimized for CPU performance on the Edge, while
maintaining speed and accuracy. The project also involved capturing
audio streams from the existing voice gateways and ensuring smooth
integration with downstream AI components for real-time processing.
This challenge highlighted the need for a flexible, resource-conscious
approach that delivered consistent and reliable results
Solutions
- Tetherfi Audio Capture Layer integrated seamlessly with the bank’s Avaya voice infrastructure, capturing audio directly from the gateway, transcoding it, and streaming separate customer and agent audio channels to the Neural Engine.
- Neural Engine powered by Tetherfi’s Speech Recognition models for efficient, chunked inference processing, delivering live transcription with optimized latency and exceptional accuracy & summarizing the entire conversation to help their agents reduce their After Call Work.
- Containerized deployment for seamless integration into the bank’s existing GenAI ecosystem, ensuring easy scalability and horizontal expansion to other business units. This intelligent design allowed real-time transcription during calls, ensuring that downstream AI applications, such as emotion analysis and agent assist modules that received timely, structured data inputs.
Benefits/Outcome
- Delivered real-time, high-accuracy transcription without reliance on costly GPU infrastructure.
- Structured transcripts, enhanced automated summarization, promise tracking, and post-call follow-up planning.
- Enabled AI-powered emotion detection and contextual agent assistance during live interactions.
- Simplified operations with a plug-and-play transcription solution fully compatible with existing voice gateway systems.
Get this Case Study Document
Please provide your name and email address for your free download.
