I was hoping to build an Agent that can accept images pasted into the chat. I like being able to use the ChatGPT app to quickly screenshot something, paste the image into chat from my clipboard, and ask it to extract the details.
It appears that the agents only currently have text input and using an email trigger is not how I wanted to code up this Agent.
Indeed, the agent is not able to parse email attachments yet. The retool team said this is in their near roadmap, but they didn't provide an ETA for this.
I guess the workaround (as for many things regarding context right now) is to upload the image to storage, and then add a tool to the agent that extract that image so that it can analyze it....
There's a lot around passing context to agents in this thread
Hi @khill-fbmc,
Confirming exactly what @MiguelOrtiz said, it is on the roadmap and I will be letting the team know to add you +1 to this FR!
For the meantime, Miguel's workaround will work. The other option would be to use a workflow to trigger an email to the model.
Passing the image from an app to a storage location that is accessible for the workflow. Not ideal but working on getting that easy drag-and-drop soon!