Uncovering the Hidden Potential of System Prompts in AI Startups
Brad Menezes, the CEO of Superblocks, a cutting-edge enterprise vibe coding startup, has a unique perspective on where the next billion-dollar startup ideas may be lurking. According to Menezes, these groundbreaking concepts are hidden within the system prompts utilized by unicorn AI startups to instruct their foundational models.
System prompts, which are detailed prompts exceeding 5,000-6,000 words, play a crucial role in guiding AI startups on how to generate their application-level AI products. Menezes views these prompts as a master class in prompt engineering, as each company tailors its system prompt to train the foundational model for specific domains and tasks.
While system prompts are not necessarily concealed, they are not always readily accessible to the public. To shed light on this valuable resource, Superblocks recently shared a file containing 19 system prompts from popular AI coding products like Windsurf, Manus, Cursor, Lovable, and Bolt in conjunction with the launch of their new enterprise coding AI agent named Clark.
Menezes’ tweet announcing the release of these system prompts went viral, catching the attention of industry heavyweights like Sam Blond and Aaron Levie. Following this buzz, Superblocks secured a $23 million Series A extension round, totaling $60 million for their innovative vibe coding tools tailored for non-developers within enterprises.
Decoding System Prompts: The Key to Unlocking AI Potential
When delving into the intricacies of system prompts, Menezes emphasizes that while the prompt itself provides a foundation for AI models, the true magic lies in prompt enrichment. This enrichment entails building a robust infrastructure around the prompt to enhance interactions with the AI model, ensuring accuracy and effectiveness in responses.
According to Menezes, there are three crucial components to analyze within system prompts: role prompting, contextual prompting, and tool use. Role prompting establishes consistency and purpose, contextual prompting offers essential context for decision-making, and tool use enables agentic tasks beyond text generation.
By studying various system prompts, Menezes gained insights into the emphasis placed on different aspects by different vibe coding platforms. Some tools prioritize fast iteration, while others focus on creating full-stack applications. Recognizing the potential for non-programmers to leverage these tools, Menezes identified opportunities to enhance security and access to enterprise data sources.
Superblocks has already attracted notable clients like Instacart and Paypaya Global, showcasing the practical applications of their vibe coding tools. Internally, the company practices what they preach by utilizing AI agents for various tasks, empowering their team to build custom tools tailored to their needs.
As Menezes continues to innovate and explore the untapped potential of system prompts, Superblocks stands at the forefront of revolutionizing enterprise coding with AI technology.
Correction: This story has been updated to clarify the recent funding round as an extension round and to reflect the updated total amount raised in the Series A.