At Fivetran, they have observed that major cloud data platforms are converging on similar features and capabilities, including vectorized SQL query engines written in C++, data "lakehouse" architectures, and Python DataFrame APIs. These features improve performance, enable ACID-compliant access to external data lakes, and allow for easier prototyping and productionization of machine learning workflows. The development of these features is a response to heightened needs for performance, scalability, and efficiency, particularly in the context of predictive modeling and machine learning. As a result, cloud data platform providers are differentiating themselves through native connectors to SaaS systems, file stores, and transactional databases, as well as offerings such as vector databases, integrations with foundation models, and other machine learning operations tools.