Superblocks launches 3.0 for AI-built enterprise apps inside AWS VPCs

Superblocks launched version 3.0 of its AI app-building platform on Monday, introducing a deployment model that runs the entire platform, including AI inference, inside a customer's Amazon Web Services (AWS) environment. The release, announced by co-founder and CEO Brad Menezes and co-founder and CTO Ran Ma, targets enterprises concerned about security risks from employee-generated AI applications, offering a managed path from prototypes to production with AWS as the boundary. Superblocks' Cloud-Prem configuration places the control plane, data plane, and development environment in the customer's AWS account, with prompts sent through Amazon Bedrock and data stored in Aurora or S3 within the customer's VPC.

Brad Menezes @bradmenezes https://x.com/bradmenezes and Ran Ma launched Superblocks 3.0 on Monday, adding a deployment model that puts its AI app-building platform, inference and generated software inside a customer's Amazon Web Services environment. Superblocks announced the release in a five-post thread on X https://x.com/bradmenezes/status/2084328970649026710 alongside what it described as a strategic partnership with AWS. https://x.com/bradmenezes/status/2084328970649026710 https://x.com/bradmenezes/status/2084328970649026710 Menezes, Superblocks' co-founder and CEO, and Ma https://www.linkedin.com/in/ranma , its co-founder and CTO, have spent much of their careers around the infrastructure that large companies use to build and operate software. The pair previously founded Apollo Health, a Y Combinator Summer 2015 company https://www.ycombinator.com/companies/apollo-health that is now inactive. Menezes later worked on application performance monitoring at Datadog, while Ma led software engineering at Confluent after an earlier role at Morgan Stanley. Menezes has said the idea for Superblocks came from watching Fortune 100 companies use Datadog to monitor internal systems for finance, support and employee onboarding. Those applications were critical to operations, even when companies treated them as secondary engineering work. Menezes told Pulse 2.0 https://pulse2.com/superblocks-profile-brad-menezes-interview/ that he and Ma saw an opening to let domain experts build that software without giving up the controls normally imposed by engineering and IT. AWS becomes the production boundary Superblocks 3.0 https://www.superblocks.com/blog/superblocks-3-0-secure-private-vibe-coding-on-aws-built-for-the-age-of-ai-cyberattacks lets employees import prototypes created with Claude, ChatGPT, Lovable, Replit or raw code, then move those applications through a company-controlled development and deployment process. Its Builder MCP also lets users modify Superblocks applications from interfaces including Claude, Cursor, ChatGPT and Slack. The AWS deployment is the center of the release. Under Superblocks' Cloud-Prem configuration, the control plane, data plane, development environment and AI inference run in the customer's AWS account. Superblocks says prompts can be sent through Amazon Bedrock using models approved by an administrator. Application data can be stored in Aurora databases or S3 buckets created inside the customer's virtual private cloud. Superblocks' deployment documentation https://docs.superblocks.com/enterprise/deployment-overview distinguishes that configuration from its cloud and hybrid products. The hybrid version keeps production data and backend execution in a customer's VPC while running the Clark AI agent in Superblocks' cloud with non-production data. Cloud-Prem keeps the entire platform inside the customer's AWS environment. That architecture targets the problem Superblocks needs enterprises to recognize: employees can already generate functional software with consumer AI products, while security teams have limited visibility into the resulting code, packages, credentials and data access. Superblocks is selling a managed route from those experiments into production, with AWS serving as the technical and procurement boundary. Superblocks had already entered the AWS Marketplace AI Agents and Tools category https://www.businesswire.com/news/home/20250716968776/en/Superblocks-Announces-Availability-of-its-Enterprise-Vibe-Coding-Platform-in-the-New-AWS-Marketplace-AI-Agents-and-Tools-Category in July 2025. Its current AWS Marketplace listing https://aws.amazon.com/marketplace/pp/prodview-kllccta3zgs2q offers a 12-month package at $149,999 before additional usage and infrastructure costs. The 3.0 release extends that relationship deeper into deployment: Superblocks wants a customer's existing AWS identity, networking, encryption and audit policies to govern employee-built applications. Security controls become the product Superblocks says every code change can pass through specialized security agents and deterministic scanners before reaching production. Those systems check for issues including SQL injection, exposed secrets, insecure data flows, authorization failures and vulnerable dependencies. Administrators can add company-specific policy agents, restrict applications to private package registries and monitor a software bill of materials for newly disclosed vulnerabilities. The release also includes Smart Router, which divides an application build into separate tasks and assigns them to different models. Superblocks says using lower-cost open models for routine coding and frontier models for planning or difficult reasoning can reduce inference costs by as much as 30%. That figure is a company estimate rather than a published customer result. Menezes made a broader claim in the launch thread, writing that Superblocks can replace "million dollar SaaS" from a single prompt. The accompanying demonstration shows Superblocks generating a CRM and importing Salesforce data. Replacing an established enterprise application would also involve migrating workflows, permissions, integrations and support responsibilities, making the claim a test of long-term operation rather than initial code generation. Superblocks is using the release to offer to build 1,000 applications free for companies that identify software they want replaced. The campaign doubles as customer research: it gives Superblocks a pipeline of expensive or unpopular SaaS workflows to reproduce while showing prospective buyers what employee-generated internal software can handle. Superblocks has reported $60 million in total funding. That includes a $23 million financing announced in May 2025 https://www.businesswire.com/news/home/20250520713435/en/Superblocks-Raises-%2423M-to-Securely-Generate-Enterprise-Apps-in-the-Era-of-AI-Vibe-Coding from Spark Capital, Kleiner Perkins, Greenoaks and Meritech Capital, following a $37 million round announced on August 11th, 2022 https://www.prnewswire.com/news-releases/superblocks-announces-37m-in-funding-to-reinvent-internal-software-development-301604330.html . The founders are spending that capital on a change in who gets to create enterprise software. Superblocks began as an internal-tool platform for developers. Superblocks 3.0 pushes the builder role into operations, finance, data and other business teams, while reserving infrastructure and policy control for IT. AWS gives Menezes and Ma a credible place to draw that line.