Paul Scanlon, a Technical Product Marketing Manager, explores the concept of "vibe coding" for AI-driven synthetic data generation, aiming to streamline development and testing workflows at Neon. Vibe coding involves using natural language to instruct AI, bypassing traditional coding details, which Scanlon sees as beneficial for creating synthetic data across various PostgreSQL databases. Despite initial challenges, such as maintaining referential integrity and managing model overloads, Scanlon experimented with both AI-only and hybrid approaches using models like Anthropic and GPT-4.5, integrating tools like faker.js to generate SQL INSERT statements. The process highlighted the complexities of maintaining data integrity at scale, suggesting that refining prompt-writing and incorporating more explicit instructions could enhance AI performance. Scanlon plans to further refine his techniques and explore educational resources to improve the outcomes of AI-driven data generation.