Call center ai · Production

AI-powered insurance call center agent assist using RAG, Amazon Transcribe, Cohere, and MongoDB Atlas Vector Search

The problem

Insurance call center agents cannot quickly locate accurate information because relevant knowledge is buried in large volumes of unstructured audio files, causing customer frustration and dissatisfaction.

Workflow diagram · grounded in source
1
Audio files stored
integration
“Storage of raw audio files: Past call recordings are stored in their original audio format”
2
Transcribe and vectorize recordings
ai_action
“Processing of the audio files with AI and analytics services (such as Amazon Transcribe Call Analytics): speech-to-text conversion, summarization of content, and vectorization”
3
Vectors stored in data store
integration
“The generated vectors and associated metadata (e.g., call timestamps, agent information) are stored in an operational data store”
4
Live customer call received
trigger
“Amazon Transcribe, which receives the audio coming from the customer's phone and converts it into text”
5
Query text vectorized
ai_action
“Cohere's embedding model, served through Amazon Bedrock, vectorizes the text coming from Transcribe”
6
Vector search retrieves FAQ
ai_action
“MongoDB Atlas Vector Search receives the query vector and returns a document that contains the most semantically similar FAQ in the database”
7
Answer shown to operator
output
“This information is finally presented to the customer service operator in text form”
Reported outcome

By converting call recordings to searchable vectors and deploying a RAG pipeline, the system enables agents to quickly access relevant information and improves operational efficiency and customer satisfaction.

Reported metrics
Information access speedquickly access relevant information and improve customer service
Operational efficiency and customer satisfactionsignificantly enhance both operational efficiency and customer satisfaction
Inquiry resolution speedaccelerate inquiry resolution
Reported stack
Amazon TranscribeAmazon Transcribe Call AnalyticsDataworkzLLMs
Source
https://www.mongodb.com/blog/post/ai-powered-call-centers-new-era-of-customer-service
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

By converting call recordings to searchable vectors and deploying a RAG pipeline, the system enables agents to quickly access relevant information and improves operational efficiency and customer satisfaction.

What tools did this team use?

Amazon Transcribe, Amazon Transcribe Call Analytics, Dataworkz, LLMs.

What results were reported?

Information access speed: quickly access relevant information and improve customer service; Operational efficiency and customer satisfaction: significantly enhance both operational efficiency and customer satisfaction; Inquiry resolution speed: accelerate inquiry resolution (source-reported, not independently verified).

How is this call center ai AI workflow structured?

Audio files stored → Transcribe and vectorize recordings → Vectors stored in data store → Live customer call received → Query text vectorized → Vector search retrieves FAQ → Answer shown to operator.

Related call center ai cases