“Every AI application needs a place to store what it 'remembers,' and that's a new database category.”

Traditional databases weren't built for similarity search across embeddings, which is why vector databases have emerged as distinct infrastructure rather than a feature bolted onto existing systems.