Trying out the assistant workflow. Pretty cool!
Feedback regarding this section: OpenAI Platform
I think this section should show how to get the last message. Seems like the most common use case. This was working:
messages = client.beta.threads.messages.list(thread_id=thread_id)
last_message = messages.data.content.text.value
Your docs aren’t open source, correct? I would have sent in the proposed addition myself, but couldn’t find your docs repo…
General first impressions of the assistant experience:
- This assistants overview page would be easier to use if it linked to the relevant API and response references throughout.
- These new deeply nested responses are hard to grok.
- The fact that they’re not subscriptable is a bit annoying. E.g.
messages['data'] causes an exception.
Downvote the page, give feedback.
The change to “not subscriptable” is more the openai python library updates that came out at the same time as the additional features. The same pydantic model based openai library was available in beta test before any details of assistants, tools, or other API revelations were released.
I wrote a demo of getting out image data from the response format not documented. It uses ‘model dumps’ after slicing the object into metadata and the actual json shown.
Funny enough, I made a big push 5 years ago to get these feedback widgets in all of my company’s docs, and here I am forgetting to use the widget myself!
Sorry, yes, I’m aware this feedback is not actually related to Assistants. Just sharing my first impression of the new response shapes.
Thanks for the demo, looks useful.