We built ORMs to make databases easier — and in doing so, we built ourselves a trap. Doctrine’s abstractions gave us a common language for SQL, but they also taught us to think in tables, not data. As modern databases evolve — document stores, graphs, vectors — those abstractions are holding us back, forcing everything into a relational mindset that no longer fits.

This talk is a wake-up call for anyone who’s ever fought Doctrine’s mapping rules, cursed a leaky abstraction, or tried to support multiple databases “transparently.” We’ll explore how our obsession with hiding complexity has stifled innovation, why we’re about to make the same mistake with AI abstractions, and how we can design better tools that embrace — rather than erase — the unique strengths of modern data systems.