Hi just a quick question. I have a social media account that uploads short videos (insta/tiktok). I’d like to have an assistant that can learn from these videos and simply talk to them.
Currently I use the chat completion API with system messages, telling the assistant what to do and what he knows but it would be cool if he could learn that by himself using the social media account.
Is this possible?
The assistants framework does not have any mechanism to connect to the internet at all.
You would need to build your own tools that the AI model can use for accessing video URLs, extracting subtitles or generating your own transcriptions from audio, using AI vision to create text about what is happening in a video from image snapshots. That would be a multi-layer task to do on-demand.
For your goal of “talk to them”, which would be to give the AI new knowledge, instead, you would likely want to obtain transcripts as text files, and then upload those as vector store documents for an Assistant. The AI then could perform searches and answer questions based on the knowledge present across all videos.
1 Like
Hi thanks a lot for the answer. I’m probably gonna build some sort of transcription tool and then give the assistant those transcriptions.
Glad to give you some ideas!