BLP
Helping businesses automate complex work, reduce manual effort, and get time back
A governed AWS foundation for an AI-native platform running 300+ agents and millions of exceptions a day
Project Overview
BLP runs 300+ AI agents for 550 enterprises. copebit built the governed AWS foundation behind it, from Control Tower landing zone to AWS Marketplace.
BLP is a Zurich-based AI-native company with more than 180 employees, combining decades of ERP expertise with AI research from Switzerland’s leading labs.Â
Trusted by more than 550 enterprises, including Migros, KRONE, Testo and FREITAG, BLP automates complex workflows across core business processes, reducing repetitive manual work and helping teams handle operational complexity more intelligently.Â
Its technology has evolved from deep expertise in ERP environments into a platform designed to make enterprise automation more reliable and easier to scale. As BLP expands across Europe and the US, the requirements behind that platform extend beyond AI capabilities alone: enterprise customers expect dependable infrastructure, strong security and compliance, and an architecture that can support continued growth. This combination of Swiss engineering, AI-native product development and business-critical enterprise workloads provides the context for BLP’s work with copebit and AWS.
The Trigger
More than 300 AI agents are at work inside the BLP platform at any given time, moving through enterprise workflows, processing exceptions and taking action across connected ERP systems. Together, they form a digital workforce operating continuously across the business processes BLP’s customers depend on.
By April 2026, BLP was already serving 450 customers. Three months later, that number had grown to more than 550.Â
The platform behind that growth needed to handleÂ
Millions of exceptions every day
with over 300 pre-built and pre-trained AI agentsÂ
and more than 30 standard ERP integrations supporting increasingly complex enterprise workloads
For BLP’s engineering team, the implication was clear: infrastructure that had worked for hundreds of customers could not remain something they built and adapted by hand as new customers arrived every week. The question was no longer whether the platform worked, but whether its foundation could keep pace with the business. BLP brought in copebit as its AWS partner to build a new, solid and scalable foundation systematically, starting the engagement under the AWS Migration Acceleration Program and creating a path from a hand-built environment towards infrastructure designed for continued enterprise growth.
“We arrived on AWS with a platform that was already live for hundreds of enterprises, so there was no room to learn slowly.”
“We arrived on AWS with a platform that was already live for hundreds of enterprises, so there was no room to learn slowly.”
“We arrived on AWS with a platform that was already live for hundreds of enterprises, so there was no room to learn slowly.”
The Road to Success
copebit built the AWS environment as a governed, repeatable platform for BLP’s next stage of scale. Rather than reproducing the existing estate service by service, the architecture establishes common patterns for infrastructure, identity, Kubernetes and AI inference, while keeping workloads and inference in European AWS regions.
A governed AWS foundation. AWS Control Tower provides the multi-account landing zone, managed as Terraform code with segmented networking, centralized backup, AWS Security Hub and Amazon Inspector. Microsoft Entra ID is federated into AWS IAM Identity Center using SAML 2.0 and SCIM, giving BLP a consistent identity and provisioning model across AWS accounts.
A processing platform built on Amazon EKS. Dozens of microservices run on Amazon EKS, providing the compute layer for BLP’s document-processing workloads.
AI infrastructure designed around the workload. High-volume OCR and reasoning run on Amazon Bedrock, with the model selected per processing step: state-of-the-art LLM models for different throughput and reasoning requirements. The inference engine moved from a fixed instance to an Auto Scaling group with Prometheus-driven scaling, allowing compute capacity to follow workload demand instead of remaining statically provisioned.
AWS as part of the commercial architecture. copebit also took BLP through its AWS Marketplace SaaS listing, including metered and contract pricing and automated private offers.
“Handling millions of exceptions a day requires a foundation that is governed, reproducible, and built to scale. A Terraform-managed landing zone, segmented accounts and per-step model choice on Amazon Bedrock in European regions mean BLP can use fast models where throughput matters and high-end models where the reasoning is hardest, without renegotiating compliance each time — and the same foundation is what lets them sell through AWS Marketplace into markets they weren’t in a year ago.”
“Handling millions of exceptions a day requires a foundation that is governed, reproducible, and built to scale. A Terraform-managed landing zone, segmented accounts and per-step model choice on Amazon Bedrock in European regions mean BLP can use fast models where throughput matters and high-end models where the reasoning is hardest, without renegotiating compliance each time — and the same foundation is what lets them sell through AWS Marketplace into markets they weren’t in a year ago.”
“Handling millions of exceptions a day requires a foundation that is governed, reproducible, and built to scale. A Terraform-managed landing zone, segmented accounts and per-step model choice on Amazon Bedrock in European regions mean BLP can use fast models where throughput matters and high-end models where the reasoning is hardest, without renegotiating compliance each time — and the same foundation is what lets them sell through AWS Marketplace into markets they weren’t in a year ago.”
The Outlook
The AWS foundation is now becoming more than the infrastructure behind BLP’s platform. It also gives the Swiss company a route to new enterprise customers, connecting the technical migration with how BLP sells, transacts and grows internationally.
From platform to route to market. BLP now transacts through AWS Marketplace, with automated private offers and its first customer orders completed through the channel. For enterprises purchasing against committed AWS spend, this provides an established procurement path to BLP’s software.
One AWS motion from migration to pipeline. The AWS Migration Acceleration Program supports the migration, AWS Marketplace provides the transaction and billing rail, and co-sell through AWS Customer Engagements (ACE) connects BLP with AWS account teams and their customers. Infrastructure and go-to-market are therefore part of the same investment rather than separate initiatives.
The migration continues. BLP is targeting feature parity with its existing estate by the end of 2026, with new customers moving onto AWS end to end. Looking forward, the engineering focus is on completing that transition and deepening the use of models through Amazon Bedrock as the platform evolves.
“Now the pattern has flipped: when we want to move faster on AWS, we pull copebit in early rather than after we hit the wall. They know our architecture as well as we do, so a decision that used to cost us weeks of investigation takes one conversation. That is what has accelerated our AWS journey — not just the build, but how quickly we can keep changing it.”
“Now the pattern has flipped: when we want to move faster on AWS, we pull copebit in early rather than after we hit the wall. They know our architecture as well as we do, so a decision that used to cost us weeks of investigation takes one conversation. That is what has accelerated our AWS journey — not just the build, but how quickly we can keep changing it.”
“Now the pattern has flipped: when we want to move faster on AWS, we pull copebit in early rather than after we hit the wall. They know our architecture as well as we do, so a decision that used to cost us weeks of investigation takes one conversation. That is what has accelerated our AWS journey — not just the build, but how quickly we can keep changing it.”
Want The Full Story?
Download the complete case study with detailed architecture insights and results.