Skip to main content
IrisAgent Private runs in your own cloud account (AWS, Azure, or GCP), inside your VPC or tenant. Everything that touches customer conversation data runs inside that boundary, including a privately hosted model.
No customer conversation data leaves your VPC, and there are no calls to public model APIs. Inference, embeddings, retrieval, and logging all run inside the boundary. Connections to your ticketing, CRM, and knowledge systems stay on your network or use private networking (private link, peering, or service endpoints), never the public internet.

Components

All of these run inside your VPC or tenant:

Deployment flow

1

Provision in your tenant

Install the stack in the cloud account, region, and VPC you choose. Nothing runs in IrisAgent cloud.
2

Connect data sources over private networking

Connect ticketing, CRM, and knowledge systems with private link, VPC peering, or service endpoints.
3

Deploy via your IAM

Deploy and operate with your own identity and access management. You control roles, permissions, and keys.
4

Tune

Configure conversation behavior, retrieval, response style, and QA rules. Validate on your own knowledge and ticket data.
5

Go live in days

Route chat, voice, and email to the deployment. Go live typically in days rather than months.

Keys and encryption

  • Customer-managed keys (KMS). You hold and rotate the encryption keys. IrisAgent never has access to them.
  • TLS in transit between channels, services, and datastores.
  • Encryption at rest with your keys for the database, vector store, object storage, and audit store.

Data flow: zero egress

Every component is inside the boundary. The public internet and third-party model APIs are outside it, with no connection.

Next steps

Email us to scope a customer-cloud deployment.