We have an app built with the service AI Applications (formerly Vertex AI Agents and 100 other names until now). It uses Vertex AI search and we have ingested some 70 docs via GCS.
The out of the box widget has a feedback feature, which allows users to rate the answers with a thumbs down and thumbs up and also provide specific reasons. Our custom app though is developed with Python and we would like to utilize this feature and reporting. However we are unable to find the documentation for the API endpoint/method that would allow us to do so.
There are many such features in the widget which could be of great help if available via the API.
Any help would be much appreciated. Thanks in advance.
To add feedback as end users browse or search your site, this documentation on the ‘add-feedback’ UserEvent type may be helpful for your use case. User Events capture all metadata related to end-user interactions.
Additionally, you can refer to this similar and relevant GDPF discussion for insights on how to implement a similar use case for widget feedback.
I reviewed both options and unfortunately not super helpful.
The question is whether the mechanism available via the widget (as noted in the discussion you highlighted) is available for developers building their own AI search interface? Since the widget is not very customizable, we prefer to develop our own frontend, however hoping that features available for the widget are also available to us.
The widget seems to use the feedback-pa endpoint which seems to be an internal internal service for Google.
Now with out of the box analytics available for AI applications, it would be nice if we could collect feedback without having to setup additional infrastructure (like databases) for it.
I would appreciate if someone from the AI applications product team could have a look at my question.