---
title: AI Apps
description: Create and configure AI-powered applications using a no-code interface to enable core DataRobot services without having to build models and evaluate their performance in DataRobot.

---

# AI Apps {: #ai-apps}


{% include 'includes/no-code-app-intro.md' %}

The following sections describe the documentation available for DataRobot No-Code AI Apps:

Topic | Describes...
----- | ------
[Create applications](create-app) | Create applications from the **Applications** tab or **Deployment** inventory.
[Manage applications](current-app) | Launch, share, duplicate, or delete applications from the **Applications** tab.
[Edit applications](edit-apps/index) | Configure your application widgets, pages, settings, and more.
[Use applications](use-apps/index) | Use a configured application to make predictions and interpret insights from your data.
[Time series applications](ts-app) | Create time series applications and consume insights in the What-if forecasting widget.

## Considerations {: #considerations}

Consider the following before deploying an application.

* The following project types _are_ supported in DataRobot applications:
    * Binary classification
    * Regression
    * Time series
    * Geospatial
    * Multiclass

* No-Code AI Apps do not support features generated by DataRobot.

* Exponentiation (i.e., `**`) is not a supported feature transformation for custom features.

* Users accessing applications via a sharing link cannot:
    * Make batch predictions from assets in the AI Catalog.
    * Submit a batch prediction to create the forecast or scenarios on the time series What-If app.

* Chart widgets display two types of data:
    * Raw data (training dataset file): Chart widgets will only display training data when the app is created from a project in the AI Catalog.
    * Prediction data: The prediction results from all single and batch predictions made in the application.

* The following is not supported when [creating an application from a Leaderboard model](create-app#from-the-leaderboard):
    * You cannot create an application from time series models on the Leaderboard&mdash;you must do it from the deployment.
    * You cannot duplicate apps that have been created from Leaderboard models.

* Organizations are limited to 200 applications. To remove this limit, contact your DataRobot representative.
* For users with _Read_ access only, Prediction Explanations must be manually computed for the model. If a user has _User_ access to the project, Prediction Explanations are automatically computed.
* While there is no limit to the number of flexible features you can specify in Optimizer applications, if the Grid Search algorithm is selected, then the grid cannot contain more than 10000 points and will return an error message if it exceeds this limit. DataRobot does not recommend using Grid Search if you are optimizing more than three features.

### Time series applications {: #time-series-applications }

* Before creating a time series application, make sure:

    * The project uses `hours`, `days`, `weeks`, `months`, `quarters`, or `years` as the time unit.
    * [Known in advance (KA)](ts-adv-opt#set-known-in-advance-ka) were set during project creation and are supported by the deployed model. Some models, for example Baseline Only models, do not support KA features even if the project is configured to use them.
    * The project is deployed to a DataRobot prediction server.
    * The deployment has an association ID and the [appropriate deployment settings configured](ts-app#configure-a-time-series-deployment).

* To include calendar events in the widget, [add a calendar file](ts-adv-opt#calendar-files) to your project and include calendar events for the timeline of the training dataset and forecasting window.
* You cannot train a model on the [_Time Series Informative Features_](ts-feature-lists#automatically-created-feature-lists) list after it has been deployed to production.