Superblocks 3.0 Integrates AI App Generation Into AWS VPCs

Amazon Web Services (AWS) and Superblocks have formed a multi-year strategic collaboration to run the Superblocks 3.0 platform natively inside customer virtual private clouds (VPCs). Released on August 3, 2026, the integration allows business teams to generate functional internal applications via AI prompts—often called vibe coding—while IT maintains strict control over data, code, and infrastructure.

Cloud-Prem Deployment Model

The Superblocks 3.0 architecture introduces a “Cloud-Prem” deployment model. This shifts the entire platform stack into the customer’s AWS environment. The control plane, data plane, inference workloads, and storage all run as fully managed services within an isolated VPC. This eliminates the need to send proprietary code or personally identifiable information (PII) to external APIs, resolving the shadow IT risks associated with unmanaged AI tools.

By shifting the application building role to business teams like operations, finance, and data, IT departments can focus strictly on governance. Financial services firm Flex utilized this architecture to deploy 70 AI-powered applications across 18 departments without compromising compliance.

AWS Native Integrations

The platform relies heavily on native AWS managed services rather than third-party infrastructure. Amazon Bedrock provides the foundational language models that power the Superblocks AI agent, known as Clark. When applications require state management, Amazon Aurora automatically spins up scale-to-zero databases. Unstructured data and generated reports route to isolated Amazon S3 buckets, and access control binds directly to existing AWS IAM policies. This approach helps reduce LLM API costs in production by leveraging established infrastructure commitments.

Smart Routing and Model Independence

A new Smart Router feature decouples the generated applications from specific model providers. The router evaluates incoming inference requests and directs them based on complexity. Simple tasks route to cost-effective open-source models, while complex logic demands frontier models. Superblocks states this dynamic routing lowers inference costs by up to 30%. By avoiding hardcoded dependencies on specific providers, organizations preserve flexibility as the model landscape changes.

Availability and Corporate Background

Superblocks 3.0 is immediately available through the AWS Marketplace. This allows organizations to draw against their existing AWS Enterprise Discount Program (EDP) commitments to procure the platform. The company, co-founded by CEO Brad Menezes and Ran Ma, previously secured $60 million in funding, including a $23 million round in May 2025 led by Spark Capital and Kleiner Perkins.

If you build internal tools for business teams, evaluate whether prompt-driven generation fits your risk profile. The shift toward VPC-bound AI platforms allows operations and finance teams to build utilities independently, provided IT configures the boundary policies first.