Misumi USA: Automating industrial component configuration and order routing with Amazon Bedrock
How Atos helped a global manufacturing leader replace manual email routing and complex product configuration with an AI-driven, agentic workflow on AWS
At a glance
Outcomes
- 14 years reduced to 2 years for web category transformation
- 2,000 emails and 3,000+ documents processed daily through automated routing workflows
- 80% classification accuracy achieved during proof-of-concept testing
- 80 sextillion product configurations supported through scalable AI-enabled processes
- Greater operational control through the transition from external providers to in-house automation capabilities
Misumi USA is a leading supplier and manufacturer of industrial components for the factory automation industry. It runs a high-volume customer communications operation, processing around 2,000+ emails every day from orders and quotes to returns and support requests.
Client Challenge
Misumi’s existing processes were creating bottlenecks across nearly every stage of the order lifecycle:
- Manual bottlenecks: The team relied heavily on manual effort to route approximately 2,000 inbound emails per day, and a significant portion of those orders were being misrouted.
- High document volume: Beyond the emails themselves, the team had to process more than 3,000 attached documents daily — all while trying to transition away from legacy third-party services toward an in-house, AI-assisted e-commerce and routing ecosystem.
- Configuration complexity: Customers navigating Misumi’s engineering-based product configurator faced a genuinely staggering level of complexity — products with 80 sextillion possible configurations — which led to high bounce rates and frequent technical error messages.
- Data silos: Researching and onboarding new web product categories took 3–6 weeks per category, largely because the process was highly manual and the necessary institutional knowledge was concentrated in just one or two people.
Taken together, these challenges meant Misumi’s growth was constrained not by demand, but by the operational capacity to process it.
Solution
Atos and Misumi designed a multi-part solution centered on generative AI and agentic automation.
- Gen AI email classification. The team used Claude 3.5 Sonnet to automatically identify orders and quote requests from inbound emails, achieving 80% accuracy in the initial phase — a major step toward eliminating manual triage.
- Automated data extraction. Bedrock Data Automation was implemented to extract critical data from the 3,000+ documents attached to emails daily, removing another manual chokepoint from the workflow.
- S3-based customer lookup. Rather than relying on constant API calls, the team built a daily data feed into an S3 database, allowing the system to automatically identify and code customers as part of the automated process.
- AI-assisted CMS. Misumi’s homegrown content management system was re-architected to use AI to identify duplicate attributes and recommend more intuitive data groupings — directly addressing the data silo and slow-category-research problem.
Technical and program details:
- Associated solution: Gen AI email classification and e-commerce transformation
- AWS services used:
- Amazon Bedrock — powering parametric search, email classification (via Claude 3.5 Sonnet), and Bedrock Data Automation
- Amazon S3 — storing customer database extracts and daily data updates
- Amazon Connect — streamlining technical support calls, with potential to integrate agentic voice AI
- AWS Lambda / API gateway — enabling real-time CAD generation and UI updates in the react-based frontend
Business Benefits
The new architecture delivered measurable improvements across efficiency, accuracy, and strategic control:
- Process efficiency: Successfully automated the routing of high-volume inbound communications, significantly reducing the friction caused by manual human intervention.
- Operational scalability: The solution targeted cutting the timeline for web category transformation from a projected 14 years down to just 2 years.
- Enhanced accuracy: Achieved 80% accuracy during the proof-of-concept phase for high-level category testing.
- In-house control: By moving away from external providers like IBM, Misumi gained the ability to build specialized logic in-house — including shipping code validation and direct, automated responses to customers.
The business case: Total cost of ownership analysis
Alongside the technical build, Atos and Misumi conducted a total cost of ownership (TCO) analysis — a comprehensive financial comparison benchmarking Misumi’s “current state” (on-premises and legacy data centers) against a “future state” running on AWS, over a three-to-five-year horizon.
The analysis began by quantifying the full cost of the current environment, including hardware capital expenditure, power, cooling and the manual labor required to maintain servers, and compared that against AWS’s operating-expense-based model. The team used tools like AWS Migration Evaluator to “right-size” resources, ensuring cloud cost estimates were grounded in actual utilization rather than over-provisioned peak capacity, while also factoring in savings available through savings plans and managed services. The study concluded that Misumi could expect a significant percentage reduction in spend, alongside a broader shift in focus — from simply “keeping the lights on” to actively driving business agility and innovation.
Lessons Learned
The TCO analysis surfaced several lessons that extended well beyond the balance sheet:
Over-provisioning is a silent budget killer. Like many organizations in this industry, Misumi had historically purchased hardware based on peak demand plus a safety buffer — often leaving servers idle as much as 80% of the time. The analysis showed that AWS’s elasticity, combined with right-sizing instances to match actual workload patterns, allows companies to stop paying for “zombie infrastructure.” The deeper lesson wasn’t just about moving to the cloud, but about fundamentally shifting the organization’s mindset from ownership to utility.
Labor and hidden facilities costs often outweigh hardware costs. IT budgeting conversations tend to focus on the sticker price of servers, but the TCO analysis revealed the much larger drain of “keeping the lights on” — rack management, cooling, power redundancy, and manual patching. Migrating to managed services showed that the real value of cloud adoption is human capital reallocation: when engineers are freed from low-value maintenance work, they can redirect that time toward higher-value innovation — new features, better customer experience — which drives more business growth than infrastructure savings alone.
On-premises costs are often opaque; the cloud demands new accountability. Misumi found that on-premises costs tend to be buried in general corporate overhead, while AWS provides granular, hourly cost data. That visibility is valuable, but it also requires a genuine “FinOps” mindset to prevent cloud waste from creeping back in over time.
The transition period itself has a cost. Because the TCO analysis spanned a three-to-five-year horizon, it also highlighted that running both the legacy and new environments simultaneously during migration is, in itself, an expensive phase that needs to be planned for.
Why Atos
Looking ahead
For Misumi, this engagement represents more than a single automation project — it’s a shift in how the organization approaches scale itself. By combining agentic AI workflows on Amazon Bedrock with a clear-eyed financial case for cloud migration, Misumi and Atos built a foundation that turns previously unmanageable complexity — 2,000 daily emails, 3,000+ daily documents, and a product catalog with 80 sextillion configurations — into a system built to grow rather than one that constrains growth.
Dive Deeper
