- DataRobot Core incorporates tools for data scientists that the company gained from its acquisition of Zepl in May 2021.
- The product offers specialized capabilities that enable users to toggle between the development of customized machine learning models and the use of AutoML for experimentation.
DataRobot’s mission is to help enterprises implement ‘Augmented AI’ by offering products that enable users to combine human intelligence with machine automation. The DataRobot AI Cloud platform facilitates customers’ ability to toggle between automation and human intervention, so that they can adjust their AI strategy as needed throughout the duration of a project’s lifecycle. To support this strategy, the company launched DataRobot Core in December 2021. The platform enables data scientists to use the tools and languages they prefer, to speed up experimentation and accelerate AI deployments.
DataRobot Core builds on the AI Cloud platform the company launched in September 2021, incorporating capabilities acquired by its May 2021 purchase of California-based Zepl. With Composable ML, AI specialists can build customized machine learning models (DataRobot Core supports Python, R, Scala, Spark and SQL), while retaining the ability to shift to AutoML mode for experimentation. Users can toggle between code-first and automated model generation, utilizing a feature that meets the requirements of staff from different departments across the organization.
The release of DataRobot Core represents a different way of tackling the same enterprise pain points. Previously, DataRobot had focused on developing automated toolsets that could meet the requirements of a broad audience (one that included data scientists and non-AI specialists). The strategy served it well. The company claims to have worked on over a million projects. Building upon this experience, it created AI Cloud for Industries, with specific products for manufacturing, banking, healthcare, and retail, which launched in November 2021. With DataRobot Core, DataRobot is expanding its portfolio to offer solutions that give data scientists greater flexibility in their approach to AI delivery. The shift makes sense. The new product caters to data scientists’ preferences to use their language of choice and enjoy the flexibility to experiment with a range of tools and platforms. This code-first, modular approach enables them to quickly move models into production and to easily scale project deployments.
Organizations face many challenges when discussing the business value of AI and determining the ways in which AI can be applied to solve practical business problems. When it comes to implementation, users face various pain points: the time it takes to get AI into production remains unacceptably long from an ROI standpoint, and there is a disconnect between data science and business outcomes, with company resources being poured into sophisticated modeling, without clear business applications. Other challenges involve the operational side of AI, with the technology still standing apart from traditional IT operations, DevOps practices, security, and governance.
In the last few years, several late-stage startups, such as DataRobot, Databricks, and Dataiku, have launched tools to help organizations deploy ML models at scale more easily. These tools have put powerful AI capabilities in the hands of line-of-business users, who are more familiar than data scientists with the business needs of an organization. At a time when AI is top-of-mind for many enterprise leaders, who are ever more aware of the growing impact that this technology will have on their industries, offering these types of tools has been an important differentiator. However, companies also need to balance this more automated approach with tools that target the specific needs of data scientists building customized models.