Support for Image Attachments in Ava
Please enable the ability to attach and analyze images within Ava, particularly for internal usage with the Archon client where customers often attach documents and screenshots to tickets. Currently, Ava does not support attachments, leading to suboptimal responses when users attempt to get help with image-based content. For example, when users tried to have Ava describe screenshots, it failed to access the attachments, while other LLMs like ChatGPT and Claude handled the same task more effectively.
Given that the LLM backend (gpt-40) already supports image analysis, implementing this feature should be feasible. This capability would significantly improve Ava’s usability, especially for clients who frequently deal with screenshot to explain their issue
-
Ofir Cohen commented
Technical side note -
Due to the requirement arises from a need within Zendesk platform, low hanging fruit here might be adding this functionality to our zendesk tool rather than supporting it as a parameter in the api call.