What challenges did Swisscom face before AI implementation?

Scale production-ready enterprise AI agents across departments while meeting strict Swiss data protection requirements. Manage secure cross-org authentication, agent orchestration, and observability for high-volume B2C customer interactions.

Source

ExploringEvidence: Low40/100

Swisscom implements enterprise agentic AI for customer support and sales using Amazon Bedrock AgentCore

Use case typeAutomotive operations automationUpdated Jun 13, 2026

Swisscom implemented Amazon Bedrock AgentCore to build and scale enterprise AI agents for customer support and sales operations. The solution uses a multi-agent architecture with MCP and Agent2Agent communication, AgentCore Runtime, Identity, and Memory, plus Strands Agents for development velocity, tracing, evaluation, and OpenTelemetry logging.

Organization: Swisscom
Industry: Tech & Comms
Location: Switzerland
Published: December 2025

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Swisscom
Provider: AWS
Maturity: Exploring
Linked source: AWS Machine Learning Blog

The solution uses a multi-agent architecture with MCP and Agent2Agent communication, AgentCore Runtime, Identity, and Memory, plus Strands Agents for development velocity, tracing, evaluation, and OpenTelemetry logging

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 4

1Customer Support
2Sales Enablement
3Agentic AI

Implemented an Amazon Bedrock AgentCore-based multi-agent architecture for personalized sales pitches and automated customer support/self-service troubleshooting.
Integrated with Swisscom identity provider and token-based least-privilege access, with AgentCore Memory for long-term insights and AgentCore Identity for access control.
Used Strands Agents framework, MCP, and Agent2Agent for agent/tool communication.

Technologies

Amazon Bedrock AgentCore Amazon Bedrock AgentCore Runtime AgentCore Identity AgentCore Memory Strands Agents SDK Model Context Protocol Agent2Agent (A2A) protocol OpenTelemetry Amazon VPC AWS Direct Connect AWS Transit Gateway

Scalable runtime optimizes cost while maintaining secure/private traffic in VPC.

Architecture

Swisscom deployed customer-facing agents as containerized runtimes on Amazon Bedrock AgentCore Runtime inside a shared VPC. The design uses AgentCore Identity for inbound and outbound authentication with Swisscom's identity provider, AgentCore Memory for long-term session insights, MCP servers and Agent2Agent (A2A) for cross-agent/tool communication, and VPC endpoints/Direct Connect/Transit Gateway for private access to internal APIs and resources. The team used Strands Agents with tracing, evaluation, and OpenTelemetry for development and observability.

Sources & evidence1

Evidence: Low40/100Evidence strength