Wand AI is a no-code platform designed to empower everyone to construct robust AI solutions uncomplicated and intuitively. The platform's journey is initiated with the assistance of Wandi, a personal assistant dedicated to the Wand platform. Wandi is designed to help users leverage the platform optimally and expedite value creation. Users can ask Wandi about the problems the platform can solve, the data required to implement General AI and more.
The Wand platform is designed to work with tabular data, which users can import by connecting their database applications or by uploading a CSV file. The platform's strength lies in its ability to learn and identify patterns and correlations between inputs and outputs using historical data. This capability allows it to accurately predict outcomes for new and unseen data, much like using past experiences to make informed decisions about the future.
The platform offers a data suggestions feature for users working with noisy data to improve data quality. This feature identifies potential data issues and suggests fixes. The platform's magic wand feature also allows users to enhance their data to make better-informed predictions. This feature enables data augmentation by summarizing it, analyzing sentiment, and more.
Once the historical data is uploaded, users must select the value to be used as the prediction key, i.e., the value they want to forecast. The platform allows users to train their models using advanced machine learning algorithms to analyze data, detect patterns, predict outcomes, and identify trends that can significantly enhance business performance.
After training, it is crucial to validate the results. The platform's global explainability function illustrates the general significance of each data parameter input for predicting outcomes. Through local explainability, users can understand the specific factors that have the most significant impact on the importance of a particular value.
Finally, users can deploy their trained model to production and start generating value. They can connect their production data using the same data source as in training or choose another one. Once the model is deployed, the result page displays predictions for each sample. Users are advised to check the prediction distribution charts and the global and local explainability to gain insights that can help address their business problems and create even more value.