> ## Documentation Index
> Fetch the complete documentation index at: https://docs.superblocks.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Enterprise deployment
Superblocks is built to integrate seamlessly into enterprise environments, with flexible deployment options that balance speed, security, and compliance.
Superblocks offers three deployment models:
1. **Cloud** – All data and AI inference run in the Superblocks Cloud
2. **Hybrid** – Production data remains in the customer VPC for deployed apps, while only non-production data is accessed by Clark in the Superblocks Cloud
3. **Cloud-Prem** – The full Superblocks platform, including AI inference, is deployed within your AWS, GCP, or Azure environment and fully managed by Superblocks
## Superblocks architecture
Superblocks is built on a modular, enterprise-grade architecture with clear separation between control, development, inference, and execution.
* **Control Plane:** Hosts the Superblocks Editor and platform services, including metadata, configuration, access control, and orchestration required to build and manage applications.
* **Clark Development Sandbox**: A secure, ephemeral development sandbox where Clark generates and modifies code. The Devbox includes access to tools, a Vite dev server, and a browser environment for automated testing.
* **Clark AI Agent**: Executes AI inference in Superblocks Cloud (Cloud and Hybrid) or in the customer's cloud environment (Cloud-Prem). Data sent to the AI model is never used for training. Model requests and responses are retained to provide the conversational chat experience.
* **Data Plane**: Executes backend logic written in SQL, Python, or JavaScript. The data plane securely proxies requests to customer APIs
* **Global Cache**: Serves deployed application frontends globally, providing low-latency access and improved performance for end users.
### Cloud architecture
In the **Cloud** deployment model, all Superblocks services are hosted in the Superblocks Cloud. Customers securely connect their SaaS applications, APIs, and databases directly to the platform. All data is encrypted in transit and at rest. Each code execution runs in an isolated, ephemeral environment, ensuring strong tenant isolation and default data segregation.
### Hybrid architecture
In the Hybrid deployment model, the Superblocks data plane is deployed within the customer’s VPC, where all backend code executes close to customer data for low latency and data residency. Production customer data remains entirely within the customer VPC. Clark operates in the Superblocks Cloud, where AI inference is performed using non-production data only.
### Cloud-Prem architecture
In the Cloud-Prem deployment model, the entire Superblocks platform, including the control plane, data plane, and Clark inference is deployed within the customer’s cloud environment(AWS, GCP, or Azure). This deployment is fully managed by Superblocks within your cloud environment. No customer data leaves the customer cloud environment, ensuring maximum data residency, isolation, and compliance.
For global enterprises with multi-region or multi-cloud environments, a single control plane is utilized, while a data plane is deployed for each region that customer data will be accessed.
## Choosing a deployment model
| Deployment Model | Primary Advantage | Clark Inference | Clark Dev Sandbox | Control Plane | Data Plane | Network Setup |
| ---------------- | ----------------------------------------------------------- | -------------------------------------- | ---------------------------------- | ---------------------------------- | ---------------------------------- | ---------------------------------------------------------------- |
| **Cloud** | Fastest setup with fully managed infrastructure | Superblocks Cloud | Superblocks Cloud | Superblocks Cloud | Superblocks Cloud | Public endpoints, IP allowlisting |
| **Hybrid** | Production data stays in-customer VPC with cloud-managed AI | Superblocks Cloud (non-prod data only) | Superblocks Cloud | Superblocks Cloud | Customer AWS, GCP or Azure Account | Data plane reaches private data sources without opening firewall |
| **Cloud-Prem** | Maximum data residency, isolation, and compliance | Customer AWS, GCP or Azure Account | Customer AWS, GCP or Azure Account | Customer AWS, GCP or Azure Account | Customer AWS, GCP or Azure Account | Full deployment in customer cloud |
**Cloud** is recommended for organizations seeking the fastest time to value with minimal infrastructure and operational overhead. This model works best when production systems can be securely accessed from the public internet or support IP allowlisting, and when strict data residency or in-VPC execution requirements are not needed.
**Hybrid** is recommended for organizations that need to keep production data and execution within their own VPC for security, compliance, or latency reasons, while still benefiting from a fully managed control plane and AI-assisted development in the cloud. This model is ideal when production systems cannot be publicly exposed, but non-production environments can safely interact with cloud-hosted services.
**Cloud-Prem** is recommended for organizations with the most stringent security, compliance, or data-residency requirements that require all application execution, AI inference, and platform services to run entirely within their own cloud environment. This model is ideal for regulated industries or environments where external connectivity is restricted, while still allowing Superblocks to fully manage and operate the platform inside the customer’s cloud boundary.