Hey Clay Team, Why is it that if I ask Chat-GPT 4o in the UI to summarize the conclusion of a YouTube video, it is able to do this easily, but if I try to do the exact same prompt using the Chat-GPT 4o integration in Clay, I run into issues. I've tested a few consecutive videos and this has happened over and over. Here is my Clay table: https://app.clay.com/workspaces/322200/tables/t_AKtrkMVpmkRh/views/gv_Fhxba7D2Yy4e
For some extra info, I have tried the prompt in both the Chat-GPT interface and the Clay table with no other context, so it's not an issue where I'm providing additional context in the native Chat-GPT UI
Hi Clay Team, just want to make sure you saw this, since I saw some of the support queries that came in earlier and later got addressed already. If you need more time to address, that's fine
Hey James, thanks for reaching out! Apologies for the delay here, will get back to you within the next hour or so about this. It's an interesting situation that I am going to reach out to a few team members about so I have an absolute answer :)
Hey there - just wanted to check in here to see if you needed anything else! Feel free to reply back here if you do.
Hey James! After testing this out with various models, including Claygent, and doing some additional research, here’s what I found: 1. Unfortunately, I wasn’t able to get ChatGPT or any other model to directly summarize a YouTube video using its API. A potential workaround could be using a YouTube Video Summarizer tool with API access, such as those available on RapidAPI or similar platforms. 2. The reason you can upload a YouTube video URL to ChatGPT 4o within the platform and get a summary without issue is that the platform has access to certain plugins. These plugins enable additional functionality, such as transcription, summarization, and web access. The standalone API doesn’t have these due to its lack of direct internet access when the API stands alone. 3. While Claygent can perform web research when the option is toggled on (even when powered by ChatGPT 4.0), this difference is related to how our model operates on the backend and what external integrations are available in each use case. Let me know if you’d like help exploring an alternative solution, setting up an API workflow, or would like more information regarding the technical details here! :)
We haven't heard back from you in a bit, so we're going to go ahead and close things out here - feel free to let us know if you still need something!