Deployment Guide
Understand Argy deployment models: SaaS, hybrid execution agents, and on-prem Kubernetes.
Argy supports multiple deployment models to match your security, compliance, and sovereignty constraints.
Deployment models
SaaS (Cloud-Managed)
- Infrastructure: Azure Kubernetes Service (AKS), EU region: West Europe.
- Agents: managed by Argy (no customer infrastructure required).
- Best for: fast start, teams without on-prem constraints.
Hybrid
- Control plane: Argy SaaS.
- Execution agents: deployed on your infrastructure to run actions close to your internal resources.
- Best for: enterprises with network/data constraints where execution must stay inside the perimeter.
On‑premises
- Full Kubernetes deployment in your environment.
- LLM Gateway can be installed locally (Enterprise + add-on) for strict sovereignty and air‑gapped environments.
- Best for: regulated industries and environments where everything must stay on-prem.
What runs where (high level)
Argy provides user interfaces (Portal, Argy Chat, Argy Code) and a platform layer (API gateway + microservices). Workflow execution happens via an execution agent (Go binary), which can be SaaS-managed or self-hosted depending on your model.
Helm charts
Argy supports Kubernetes deployments via Helm with centralized configuration in values.yaml, including:
- Replicas per service (default: 2)
- CPU/memory resources
- Ingress (NGINX) with SSL
- Optional Network Policies
- Optional OpenTelemetry Collector
- Feature flags per environment
Next step: review the Security Model and the LLM Gateway for governance and audit details.