This page will be continually updated with the most common issues encountered by users during our beta release.
Updated July 25, 2023
Olympia is known to currently have display issues or simply not work on some versions of the following browsers:
Please use Chrome or Safari in the meantime.
If you’re trying to get long output out of the writers such as Calvin or Leticia, make sure that their conversation settings are set to long output (they are not by default.)
To change the settings, click on the settings icon at the top right of the app.
A settings dialog will pop up, where you can adjust the settings for that current conversation.
We call the unwanted interruption of responses stalling. It occasionally happens when the AI gets confused about what you’re asking it to do, or gets overwhelmed by the size of the conversation transcript or simply doesn’t want to comply, due to a number of factors.
Calvin needed some prodding…
What to do in this situation? Just type continue or … and they’ll usually pick right back up where they left off. It’s like a gentle nudge to say “Hey, weren’t you saying something?”
<aside> 😅 Make sure to use precise language if your team member won’t stop stalling! Sometimes a user asks the AI to “send the content now” and the AI responds, “ok, I’m sending the content now” and then doesn’t do anything else. If you continue saying, please send the content, it will continue saying that it will send the content.
The problem is the use of the word “send” which is incorrectly interpreted to mean to send via email or other channels, which do not exist. This error is a form of hallucination that at the moment is very difficult to prevent with GPT models, although we are gradually making them happen less often.
The solution is to use the words respond with or output to make it clear that you want the AI to reply in the chat now.
</aside>