Axonius, an asset intelligence platform, is integrating AI agents into its SaaS offering on AWS, leveraging Amazon Bedrock AgentCore to manage multi-tenant environments securely. The platform helps security and IT teams prioritize risks and coordinate fixes by reconciling data from over 1,400 systems into a single source of truth. Axonius runs its software as a service (SaaS) infrastructure on AWS, managing hundreds of isolated customer environments. The integration allows for secure AI agent deployments while maintaining tenant isolation and data privacy. Axonius opted for a silo deployment model, where each customer workload resides in a dedicated Amazon Virtual Private Cloud (Amazon VPC). This approach ensures tenant-specific data access and security, aligning with the company’s existing infrastructure and operational practices.
Axonius faced several key requirements when integrating AI agents, including tenant isolation, identity integration, cost tracking, and secure API access. The company needed to ensure that an agent serving one customer is scoped exclusively to that customer’s data. Additionally, Axonius required secure access to tenant-specific APIs and lifecycle management within its Continuous Delivery (CD) workflow. The platform also needed observability tools to track large fleets of agents, issue alarms for issues, and provide tracing capabilities for debugging. These requirements shaped Axonius’s approach to deploying AI agents while maintaining its silo model.
Axonius evaluated several deployment options, including a pool model with JWT-based tenant routing and a bridge model with gateway-enforced tool isolation. The pool model uses a single runtime with JWT-based tenant routing, offering operational simplicity and rapid onboarding for new customers. However, it relies on application-level controls for tenant separation. The bridge model combines a shared runtime with a gateway that enforces tenant boundaries through policy and Lambda interceptors. This approach provides layered infrastructure enforcement and centralized governance for tenant policies. Axonius ultimately chose a model that aligns with its existing infrastructure and operational needs, ensuring secure, scalable AI agent deployments.
Source: awsml
