Vaadin Add-on that displays a chat assistant floating window using Material UI's FAB and Vaadin web components.
- Send messages from the user or programmatically, and listen for messages written by the user.
- Toggle the chat window open/closed, or open it as a full-screen dialog on mobile.
- Markdown rendering, lazy loading via a
DataProvider, and streaming ("generative") answers. - Customizable floating action button (FAB): icon, size, color (
FabVarianttheme variants), corner position and margin, draggable or fixed, and viewport- or container-anchored placement. - Resizable chat window with eight drag handles, optional resize direction indicators, and configurable initial size with min/max bounds.
- Responsive desktop/mobile modes with optional breakpoint-based auto-switching, plus listeners for mode changes and for the chat window crossing a size threshold.
- A fluent
ChatAssistant.builder()to configure everything declaratively.
Supported versions: Vaadin 24-25 (add-on version 4.x)
Add the following dependencies in your pom.xml file:
<dependency>
<groupId>org.vaadin.addons.flowingcode</groupId>
<artifactId>chat-assistant-addon</artifactId>
<version>X.Y.Z</version>
</dependency>Release versions since 5.0.1 are available from Maven Central repository. Earlier versions are available from the Vaadin add-on repository. For SNAPSHOT versions see here.
- git clone repository
- mvn clean install jetty:run
To see the demo, navigate to http://localhost:8080/
See here
The issues for this add-on are tracked on its github.com page. All bug reports and feature requests are appreciated.
Contributions are welcome. There are two primary ways you can contribute: by reporting issues or by submitting code changes through pull requests. To ensure a smooth and effective process for everyone, please follow the guidelines below for the type of contribution you are making.
Creating an issue is a highly valuable contribution. If you've found a bug or have an idea for a new feature, this is the place to start.
- Before creating an issue, please check the existing issues to see if your topic is already being discussed.
- If not, create a new issue, choosing the right option: "Bug Report" or "Feature Request". Try to keep the scope minimal but as detailed as possible.
A Note on Bug Reports
Please complete all the requested fields to the best of your ability. Each piece of information, like the environment versions and a clear description, helps us understand the context of the issue.
While all details are important, the minimal, reproducible example is the most critical part of your report. It's essential because it removes ambiguity and allows our team to observe the problem firsthand, exactly as you are experiencing it.
As a first step, please refer to our Development Conventions page to find information about Conventional Commits & Code Style requirements.
Then, follow these steps for creating a contribution:
- Fork this project.
- Create an issue to this project about the contribution (bug or feature) if there is no such issue about it already. Try to keep the scope minimal.
- Develop and test the fix or functionality carefully. Only include minimum amount of code needed to fix the issue.
- For commit message, use Conventional Commits to describe your change.
- Send a pull request for the original project.
- Comment on the original issue that you have implemented a fix for it.
This add-on is distributed under Apache License 2.0. For license terms, see LICENSE.txt.
Chat Assistant Add-on is written by Flowing Code S.A.
ChatAssistant<T extends Message> renders a floating action button (FAB) that opens a chat window.
Add it to any layout and send messages to it; it shows a FAB in the bottom-right corner by default.
ChatAssistant<Message> chatAssistant = new ChatAssistant<>();
add(chatAssistant);
// Send a message programmatically (e.g. from the assistant).
chatAssistant.sendMessage(Message.builder()
.name("Assistant")
.content("Hello, I am here to assist you")
.messageTime(LocalDateTime.now())
.build());
// React to messages typed by the user.
chatAssistant.setSubmitListener(ev ->
chatAssistant.sendMessage(Message.builder()
.name("User")
.content(ev.getValue())
.messageTime(LocalDateTime.now())
.build()));
// Open or close the window programmatically.
chatAssistant.setOpened(true);Use ChatAssistant.builder() to configure the FAB and window declaratively:
ChatAssistant<Message> chatAssistant = ChatAssistant.<Message>builder()
.fabIcon(new SvgIcon("icons/my-icon.svg")) // custom FAB icon (defaults to a chatbot icon)
.defaultFabPosition(FabPosition.BOTTOM_LEFT)
.fabMovable(true) // allow dragging the FAB
.resizable(true) // allow resizing the window
.markdownEnabled(true) // render message content as Markdown
.build();Everything in the builder also has a setter, so the FAB and window can be reconfigured at runtime.
chatAssistant.setFabPosition(FabPosition.TOP_RIGHT); // move it (and set the reset corner)
chatAssistant.addFabThemeVariants(FabVariant.LARGE); // grow the FAB (FabVariant.SMALL/LARGE)
chatAssistant.addFabThemeVariants(FabVariant.LUMO_CONTRAST); // recolor it (color variants)
chatAssistant.setResizeIndicatorsVisible(true); // show resize-direction hintschatAssistant.setWindowWidth("400px");
chatAssistant.setWindowHeight("500px");
chatAssistant.setWindowMinWidth(300); // bounds honored on open and while resizing
chatAssistant.setWindowMaxWidth(700);In MOBILE mode the chat opens as a full-screen dialog. Set a breakpoint to switch automatically
between desktop (anchored popover) and mobile as the viewport crosses it:
ChatAssistant<Message> chatAssistant = ChatAssistant.<Message>builder()
.mobileBreakpoint(768) // auto-switch to mobile below 768px (auto-switching is opt-in)
.build();
chatAssistant.addModeChangedListener(ev -> Notification.show("Now in " + ev.getMode() + " mode"));
chatAssistant.setMode(ChatAssistantMode.MOBILE); // or switch manuallyBy default, Vaadin Flow only includes com/vaadin/flow/component to be always scanned for UI components and views. For this reason, the add-on might need to be allowed in order to display correctly.
To do so, just add com.flowingcode to the vaadin.allowed-packages property in src/main/resources/application.properties, like:
vaadin.allowed-packages = com.vaadin,org.vaadin,dev.hilla,com.flowingcode
More information on Spring scanning configuration here.