Brad Menezes, CEO of Superblocks, has sparked industry conversation by publicly sharing system prompts from 19 leading AI coding tools alongside the launch of Clark, his company's new enterprise AI agent. This unprecedented transparency offers a rare glimpse into the hidden mechanics powering today's most advanced coding assistants.

The Hidden Language of AI When Superblocks released Clark, they didn't just introduce another coding assistant—they published what Menezes calls "a public course in LLM prompt engineering." The document reveals the exact instructions shaping behavior in popular tools like Cursor, Replit, and Devin. "Each company trains the same base model to become a specialist through carefully crafted prompts," Menezes explained to TechCrunch.

Beyond the Prompt: The Real Magic While revealing these prompts generated buzz, Menezes insists they represent just 20% of what makes AI tools effective. "The remaining 80% lies in how you attach auxiliary instructions and refine outputs," he notes. Superblocks breaks down their approach into three layers:

• Role Prompts define an AI's identity (e.g., "You are Devin, a code genius...")
• Situation Prompts establish boundaries (like Cursor's rule against revealing tool names)
• Tool Invocation Prompts trigger external functions from databases to shell commands

Shifting Focus to Enterprise Needs Fresh off a $23 million Series A round (totaling $60 million), Superblocks positions Clark differently—targeting non-technical teams needing custom internal tools. In one experiment, sales staff built CRM assistants while ops teams created performance dashboards—all without engineering support. "We're proving departments can build rather than buy their solutions," says Menezes.

As AI standardization grows, Menezes believes mastery of prompt engineering will separate winners from also-rans. "These prompts aren't just instructions—they're cultural DNA encoding each startup's unique value." With Clark, Superblocks bets that democratizing this knowledge while simplifying enterprise tool creation could redefine workplace automation.

Key Points

1. Superblocks disclosed system prompts from 19 major AI coding tools alongside Clark's launch
2. Prompt engineering involves role definition, situational rules, and tool activation protocols
3. The company secured $60 million total funding to target non-technical enterprise users
4. Early tests show departments successfully building custom tools without developer involvement
5. Effective prompt strategies may determine competitive advantage in maturing AI markets