Nuxeo has launched Nuxeo Insight Cloud, an AI platform for enterprises that enables nontechnical users to use and train machine learning models with their own specific data sets.
According to Nuxeo, using business-specific data is essential in effective search, workflow and other activities in content-centric business applications. When designing Nuxeo Insight Cloud, Nuxeo aimed to create an AI service that wasn't generic and could be trained with each enterprise's own data.
This, according to Nuxeo, enables enterprises to use AI models that are more accurate and relevant because the models automatically generate data values to enrich content, drive workflow and increase intelligence about information.
Nuxeo Insight Cloud determines the key information in users' data and enables them to classify and find information. It also extracts automated metadata from documents and media assets and then enriches the metadata automatically based on a business's system or information.
Then, after Insight Cloud enriches the data, it can be applied to automate image capture, launch workflows or associate new content with pending tasks or work assignments. It also learns and improves prediction accuracy over time and enables users to retrain it over time as more content and data are added.
According to Nuxeo, Insight Cloud is a three-step process. First, enterprises can train the platform by choosing what users want to be predicted for what training data the Nuxeo AI should be using. Then, Nuxeo Insight Cloud acts as a virtual assistant and learns the specifics of the data and suggesting values for users. Finally, users can accept or reject recommendations for Insight Cloud to refine its suggestions so that it can learn more about what users prefer.
Nuxeo Insight Cloud is powered by Nuxeo Platform, which provides two additional AI capabilities. The first is a service that uses Nuxeo AI to describe content in systems such as Box and Dropbox, as well as in legacy enterprise content management repositories. Second, all machine learning models are versioned and provide an audit trail for machine-generated data values.