Customer Service Automation · Document Processing
Misumi USA needed to handle about 2,000 inbound emails and more than 3,000 attached documents each day, but manual routing and document handling caused frequent misrouting, slowed order and quote processing, and made it hard to move away from legacy third-party services. Customers also struggled with an engineering-based configurator that supports 80 sextillion possible product combinations, while onboarding new web product categories took 3–6 weeks because critical knowledge was siloed.
Reported impact
Misumi automated high-volume email routing and document handling, achieved 80% classification accuracy in proof-of-concept testing, reduced the projected web category transformation timeline from 14 years to 2 years, and gained greater operational control by bringing specialized automation capabilities in house.
Atos and Misumi designed a multi-part AI architecture around customer communications and e-commerce operations. Inbound emails and quote or order requests are classified with Claude 3.5 Sonnet on Amazon Bedrock so work can be routed automatically instead of triaged by hand. Bedrock Data Automation extracts key fields from more than 3,000 attached documents per day. A daily customer-data feed lands in Amazon S3 so the workflow can identify and code customers without constant API calls. Misumi's homegrown content management system was re-architected to use AI for duplicate-attribute detection and better data grouping, while Amazon Connect supports technical-support interactions and AWS Lambda / API gateway enables real-time CAD generation and UI updates in the React-based frontend. Together these components turn high-volume manual routing, extraction, and configuration work into agentic workflows.
Customer emails and attachments enter Amazon Bedrock workflows, where Claude 3.5 Sonnet classifies requests and Bedrock Data Automation extracts document data. The process enriches requests with daily customer records stored in Amazon S3, then routes work into Misumi's in-house order, quote, CMS, and frontend experiences, while Amazon Connect and AWS Lambda / API gateway handle support interactions and real-time UI actions.
"Bedrock Data Automation was implemented to extract critical data from the 3,000+ documents attached to emails daily"
"Customers navigating Misumi's engineering-based product configurator faced ... 80 sextillion possible configurations"
"The solution targeted cutting the timeline for web category transformation from a projected 14 years down to just 2 years."
"The team used Claude 3.5 Sonnet to automatically identify orders and quote requests from inbound emails, achieving 80% accuracy in the initial phase"
"the team built a daily data feed into an S3 database, allowing the system to automatically identify and code customers"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
The Organization JSON-LD lists numberOfEmployees value 400; the standard mapping classifies 250-999 employees as Mid-market.
Related
Microsoft Sample Nov 14, 2024 Legrand Legrand, a leader in electrical and digital building infrastructures, powers everyday life by delivering essential products, from lighting controls to wiring devices. But with more than 300,000 products across 90… Azure AI Search · Azure Data Lake · Azure Data Lake Storage Microsoft Sample Sep 30, 2024 elunic But there's still a long way to go before this vision of the future is realized. Although machines already deliver several terabytes of data every day, the potential within that data isn't yet being exploited. "Some of… This solution leads to increased productivity for customers by saving 15 minutes per request. It also enables a… Azure Kubernetes Service · Azure OpenAI Service 
AWS Sample Sep 11, 2024 Georgia-Pacific The lack of a centralized, readily accessible knowledge base often leads to reduced machine productivity, prolonged machine downtime, and higher troubleshooting and repair costs. Making phone calls to experts and… Accelerated operator onboarding Amazon Bedrock · Amazon Kinesis 
