Superblocks Partners With Databricks Days After Launching Clark

Superblocks didn’t wait long to make its first enterprise move. Just days after launching Clark, its AI-native platform for internal app development, the company was announced as the lead partner for Databricks Apps at the 2025 Data + AI Summit.

Today is a BIG day!
I’m thrilled to announce a strategic partnership between @databricks and @superblocks that debuted live this morning during the #DataAISummit keynote! This marks a major leap forward in how enterprises usher in an era of secure “enterprise vibe coding” to… pic.twitter.com/SVCxrISLtq

— Brad Menezes (@bradmenezes) June 11, 2025

The partnership allows Superblocks to integrate directly with Databricks’ Data Intelligence Platform. Clark users can connect to Databricks SQL, Mosaic AI, Lakehouse data, and enforce permissions using Unity Catalog all while deploying apps inside Databricks’ secure perimeter or within a customer’s private cloud using Superblocks’ on-prem agent.

From Low-Code to AI-Native

Superblocks originally launched as a low-code platform focused on helping engineering and IT teams build internal tools. But as generative AI adoption surged, the company made a full transition. CEO Brad Menezes and co-founder Ran Zhou shifted their roadmap entirely, rebuilding the backend to store apps as raw, auditable code and equipping the platform with AI-native capabilities.

The result was Clark, a generative AI agent that enables not only developers, but also IT and operations teams, to build internal applications from natural language. Clark supports three modes of development: prompt-to-app via AI, drag-and-drop editing via a visual builder, and full extensibility through IDEs like Cursor, Windsurf, and VS Code.

I’m excited to introduce Clark, the first AI Agent to build internal enterprise apps.

We’ve raised $60M, including a fresh $23M from @sparkcapital @kleinerperkins, @MeritechCapital and Greenoaks.

Unlike consumer vibe coding tools like Lovable, Replit and Bolt that only generate… pic.twitter.com/zVlnq8BeCc

— Brad Menezes (@bradmenezes) May 27, 2025

System Prompts as Strategic Differentiators

To coincide with the Clark launch, Menezes released a file compiling 19 system prompts used by popular AI coding agents, including Cursor, Windsurf, Devin, Replit, and Manus. The release drew widespread attention on social media, with nearly 2 million views and engagement from prominent investors and operators.

Menezes said the prompts offer insight into how companies use the same foundational models in fundamentally different ways. According to him, the base system prompt is only about 20% of the real architecture, the remaining 80% lies in “prompt enrichment,” or the infrastructure that wraps around each LLM call: role prompting, contextual injection, tool instruction, accuracy checks, and model behavior enforcement.

“You basically have to speak as if you would to a human co-worker,” Menezes explained. “And the instructions have to be perfect.”

Clark leverages this approach to build governed, agentic workflows within enterprise constraints.

Not Built for Hacky Prototypes

In the months leading up to Clark, tools like Bolt, Replit, Lovable, and v0 have drawn interest from individual builders for their speed and simplicity. But Superblocks doesn’t see them as competitors. Those tools, while good for prototyping, don’t support structured deployment, enterprise-grade data integration, or compliance needs.

Menezes pointed out that platforms like Loveable and Bolt emphasize fast iteration, while others like Replit, Codex, and Manus are better at generating full-stack applications. But in his view, these outputs often remain raw code and aren’t suited for direct enterprise use.

“The most innovative enterprises are not just reducing costs with AI,” he said. “They’re unlocking 100x more opportunities to automate operations, streamline workflows, and drive revenue.”

Clark is positioned specifically for production-grade internal software. The platform includes SSO, RBAC, brand design systems, on-premise deployment options, and integrations with tools like Salesforce. Customers include Instacart and Papaya Global.

Internally, Superblocks enforces its own usage. Engineers are required to focus only on the product. Business and sales teams build their own internal tools using Clark, from CRM analytics to workload distribution apps.

Clark on Databricks

With Databricks, Clark users can generate apps directly on top of enterprise data systems. The integration supports:

• Natural language app generation on Databricks SQL, Mosaic AI, and Lakehouse
• Secure deployment governed by Unity Catalog
• End-user role-based access controls
• Visual and code-based editing of generated apps

As part of the partnership, enterprises can choose to host apps inside Databricks Apps or within Superblocks, backed by an on-prem agent for data isolation.

Databricks VP of Product Management Shanku Niyogi noted: “Our customers have invested in building a data and AI foundation in Databricks, and are increasingly turning to applications to help democratize those investments across their organization. With Superblocks, they’re able to accelerate the development of these data and AI applications, without sacrificing security and governance.”

Superblocks has raised $60 million to date, including a $23 million Series A extension led by Kleiner Perkins, Spark Capital, Green Oaks, and Meritech. Menezes, who previously held roles at Datadog and Yelp, attributes some of that success to early network credibility and tight investor storytelling.

“The key is capturing the essence of what you are doing in 15 to 20 slides,” he said. “Storytelling is everything.”