Service and Support
H2O.ai provides good customer service, particularly praised for being accessible and exceptional due to its smaller size. Users often engage with community forums for support and some avoid technical support due to the application's clarity. When used, technical support is rated highly. Experiences with the tech team in Seattle are positive, and no recent issues have been reported. H2O.ai’s support is well-regarded, available, and valued by those who have interacted with them.
Deployment
Users found H2O.ai's initial setup to be easy and straightforward through pip, JAR download, or R install.packages. Many found the process to be quick, although it took a few hours for those without expertise. They rated the experience highly, often comparing it favorably to platforms like DataIQ, which were considered more complicated. Integration and configuration were praised with users giving high scores, ranging from eight to nine out of ten.
Scalability
H2O.ai’s scalability performs smoothly for diverse user requirements, supporting both small and large data environments effectively. It accommodates intensive computational needs, making it suitable for extensive enterprise use cases. Its ability to handle parallel executions, particularly with Spark, enhances execution efficiency. However, there may be limitations in handling evolving use cases due to the varying nature of generative AI. Despite this, H2O.ai easily aligns with infrastructure specifications, ensuring scalability for different business sizes.
Stability
Users encountered minimal issues with H2O.ai while working on prototypes and found workarounds for bugs. It showed impressive stability in trials, especially with small to medium datasets. Larger datasets presented some performance challenges, as is common with any platform. Some teams noted they hadn't stressed the platform significantly yet but experienced very stable performance during preliminary evaluations, indicating H2O.ai handles early-phase testing effectively and delivers outstanding results in proof-of-concept stages.
