Entities in the Wand AI platform
- Solution – solution is an entity that represents a business problem you would like to solve using the Wand Platform. Solution is the highest entity in our hierarchy. Examples for solutions can be Customer Retention, KYC, and more. A solution consists of several AI tasks that are related to each other. In the example of Customer Retention, AI tasks could be Customer Churn Risk and Customer Satisfaction Estimation. Both tasks can help us to improve our customer retention. We created the solution level in order to simplify sharing AI tasks that are related, utilizing the same data source(s), and combining multiple AI tasks together. In the simplest way, you can think about solutions as a folder that contains related AI tasks, exactly as you have in your computer. You can create a solution for any department or any business problem you would like to solve with the Wand Platform.
- AI Task – AI task is an entity that represents a specific business question you have. For example, which customers are at risk of churning in the following 30 days. You can provide this AI task with a meaningful name such as “Churn Risk Estimation”. An AI task consists of experiments. Each experiment’s goal is to answer your question with the provided data and settings you bring. You can have multiple experiments within each AI task – to try to bring different types of data or make different data transformations in order to answer the same question and see which one works best. The first experiment in an AI task will be created automatically.
- Experiments – Experiment is an entity that represents all the steps in the AI lifecycle. Each experiment must have the same data schema after all (optional) transformations. Each experiment consists of several environments – Training, Staging (optional), and Production. Please see right below the information about them.
- Environment – an environment is an entity that represents a specific step in the AI lifecycle – training step, staging step, and production step.
- Training environment – where you bring your training data in order to create the AI model – learning from historical information, to create a model that can predict the answer for your question. The output of the training environment is an AI model. Each run of the training environment will create a new AI model. You can run the training environment multiple times and create multiple models (e.g. by bringing more data (in the same structure) and running the environment again).
- Staging environment (optional) – which you deploy your model for testing purposes. Staging environment can be used to try out the AI task on small amounts of users, or on internal users, before moving to full production. Staging environment has the same properties as the production environment, to enable you to simulate real production. A staging environment cannot exist before you create the model in your training environment. The output of the staging environment runs as a prediction for your data. Each run of the staging environment will create new predictions for your data. You can make predictions in real time, on a schedule, or on-demand.
- Production environment cannot exist before you create the model in your training environment, but can exist without a staging environment, which is optional. The output of the production environment runs as a prediction for your data. Each run of the production environment will create new predictions for your data. You can make predictions in real time, on a schedule, or on-demand.
- Runs – runs are the action of running each environment again to create the relevant output.