Axonius Implements Secure Multi-Tenant AI Agents Using Amazon Bedrock AgentCore

Axonius, a cybersecurity SaaS platform, has integrated AI agents into its service architecture to enhance user experience and automate manual tasks. To manage the complexities of multi-tenant deployment, Axonius adopted Amazon Bedrock AgentCore, a platform designed for building, connecting, and optimizing AI agents at scale.

The company chose a silo deployment model, where each customer operates within a dedicated Amazon Virtual Private Cloud (VPC), ensuring strict tenant isolation. This approach aligns with Axonius's security requirements, as it handles sensitive customer data and needs to prevent cross-tenant data access.

Axonius's architecture involves deploying a dedicated AgentCore runtime for each customer, with each runtime hosted on an isolated microVM within the customer's VPC. This setup ensures complete separation of workloads and data, with network traffic confined within the customer's private network. The runtime connects to the customer's internal systems via private network interfaces, maintaining data privacy and security.

The deployment leverages AWS services such as Amazon ECR for container storage, Amazon Bedrock for foundation models, Bedrock Knowledge Bases for data retrieval, and Amazon CloudWatch for observability and cost management. The architecture also utilizes AWS VPC Lattice for private connectivity, reducing networking complexity and costs.

This implementation allows Axonius to efficiently scale its AI agent deployment, with automated provisioning and teardown facilitated by AWS CloudFormation. The use of session isolation in AgentCore runtime ensures that each user session is securely contained, and the integration of guardrails maintains response safety and scope.

Axonius's approach demonstrates how SaaS providers can leverage Amazon Bedrock AgentCore to deploy secure, scalable, and manageable multi-tenant AI agents, significantly reducing development time and operational overhead. The company's solution exemplifies best practices in tenant isolation, private networking, and cost governance in AI SaaS architectures.