This article describes known limitations when executing Flows through WhatsApp, Slack, and Microsoft Teams.
While Flows typically perform as expected during testing in the Workspace (via the "Run Test" feature), some controls behave differently when integrated with external chat platforms or even within the Workspace itself.
The goal of this article is to ensure you are aware of these limitations when creating and deploying Flows in production environments.
Flows support interactions with the Chat with Data Assistant, allowing the exchange of messages and visual assets such as charts or images.
However, an issue has been identified specifically with images uploaded directly from user devices when using WhatsApp integrations.
Images that originate from local uploads (desktop or mobile) may not be displayed correctly in WhatsApp chats. In contrast, images referenced via public URLs are rendered as expected.
This limitation is not related to the Chat with Data Assistant itself, but rather to how WhatsApp handles image rendering from temporary or local file sources.
Recommendations:
- Prefer hosted image URLs instead of direct file uploads when using images in Flows integrated with WhatsApp.
- Store images in publicly accessible services (e.g., AWS S3, Google Cloud Storage, or similar).
- Avoid workflows that depend on real-time image uploads from user devices.
You can design interactive experiences using various input components in Flows.
However, a limitation affects the Dropdown interaction when executing Flows from the Workspace module.
In this environment, Dropdown components cannot be properly tested or validated, which impacts the development workflow.
This limitation is specific to the Workspace and does not necessarily affect external integrations such as WhatsApp, Slack, or Microsoft Teams, where behavior may differ.
Recommendations:
- Avoid using Dropdown components in Flows that need to be tested within the Workspace.
- Use alternative interaction types such as Text Input or Buttons for testing purposes.
- Do not rely on Quick Reply as a fallback within Workspace, as it also presents limitations in this environment.
The Quick Reply component allows you to present selectable options within a Flow, enabling dynamic and interactive branching based on user input.
However, there is a known limitation affecting the Quick Reply feature when used in WhatsApp and in the Workspace module.
In the Workspace, this limitation is related to how sessions are initialized. The Workspace does not emit a session start event when a user opens a conversation. As a result, the welcome message is not automatically triggered, and Quick Reply options are not rendered on screen.
This creates a mismatch between what the system expects and what the user sees:
- From the system perspective, once the user sends any input (e.g., “Hola”), the Flow state machine assumes that the user is responding to a Quick Reply and expects a numeric input (e.g., 1, 2, 3, ... n).
- From the user perspective, no Quick Reply options have been displayed yet, so they are unaware that they should respond with a number.
As a consequence, any non-numeric input will result in an error, since it does not match the expected Quick Reply format.
Additionally, in WhatsApp, Quick Reply interactions may not behave as expected, as the Flow can automatically advance without waiting for user input.
At the moment, there is no clear evidence that this issue affects Slack or Microsoft Teams in the same way, although behavior may vary depending on the integration.
Recommendations:
- Avoid using Quick Reply as the initial interaction in a Flow.
- Ensure that a clear text prompt is displayed before expecting structured input.
- Consider using Text Input or Button interactions as alternatives when working within Workspace.
- Validate behavior carefully in each target integration environment.
- These limitations are being tracked and updated regularly by the Globant Enterprise AI team.
- It is recommended to review this document before building new Flows to ensure compatibility across platforms.
- Future releases may include fixes or improved handling for affected components.