Skip to main content

We have recently migrated over to Claude AI from OpenAI. The premise of the project is a ChatBot, that needs to have access to data that can change on a daily basis. 

 Overall, Claude has performed much better for us. However the one aspect that we are having a hard time figuring out is the trainable data. For example in the Zapier to OpenAI/ChatGPT integration, Zapier has these inputs; 

 


Past this, OpenAI allows you to Vector Store data directly, before a Zap even occurs. 

Claude has relatively a simple Zapier integration. It does not allow for new data sources to be input; 

 

It also does not allow you the option to work with a specific project within Claude that might already have data integrated. 

A workaround seems to be to have data stored, and then hydrated by Claude, but it's only a temporary fix. 

Has anyone found a workaround, or does anyone have any ideas? Trying to avoid a massive python undertaking. 

 

Hi @ECC-MN, thanks so much for joining our Community! 🎉

I did some checking and you’re right I couldn’t see any option to add new data source files or select a past project either. I’d suggest contacting our Support team to submit feature requests for the ability to do that using the form here: https://zapier.com/app/get-help

In terms of a workaround the Vision (Images) field appears to accept image files so I wonder if you might be able to create an image of the data then try uploading it to see if it’s able to read it. That said, if there’s a large amount of data in the file then that’s likely not going to be a viable option.

Other than that, if Anthropic’s API allows file upload via and accepts basic authentication methods, then perhaps you could try using a Webhooks action to make a request to their API and upload the file that way.

Hope that helps to get you pointed in the right direction. If you run into any issues or have further questions do let us know! 🙂