The AI responses generated by the automation plugin do not notify the user when the quota is exhausted; instead, they display an error code

Contribute Bug
,

AI-generated summary

The discussion highlights a bug in the automation plugin where AI responses do not notify users when their quota is exhausted, instead showing error codes. kuaza explains that while the system should inform users via notifications or messages when a paid model’s quota is depleted, the automation plugin fails to do so, leaving users unaware. They suggest using a free model as a fallback and providing clear user alerts, including on-screen messages or emails, to guide users toward upgrading or exploring alternatives. Falco acknowledges the need for better quota exposure but notes the complexity due to asynchronous automation routines. kuaza emphasizes the importance of real-time notifications for enhancing user experience and community engagement.

1

I have developed an automation and configured it to respond using an artificial intelligence persona triggered by a specific tag.

We are facing an issue: when a quota is set for the AI and it is exhausted, it does not respond automatically. The error codes indicate that the quota has been depleted.

The expected behavior is: if the user’s AI quota is exhausted, the system should continue to generate responses using a designated free model (which could be a general AI selected for this purpose), but the response should also clearly indicate that the limit for the paid model has been reached, or alternatively, notify the user that their quota has been exhausted and that they will receive services from the free models until a new quota is added.

In other words, when a question is asked to a paid model and the quota is depleted, the system should inform the user in some manner. When this is done via a personalized message, the user is informed. However, when responses are generated automatically through the automation plugin, the ‘quota exhausted’ message appears in the error logs, and the user remains uninformed.

Error responding to: /t/xxxxxxxx-1758566266213/1500/1 in LlmPersonaTriage.handle : LlmQuotaUsage::QuotaExceededError : ..........

2

We do indeed need better ways to expose quotas being exhausted, but given the nature of async in automation routines that will be trickier.

1 Like
3

Expected:

  • Users should be informed immediately through notifications, on-screen messages, or, ideally, via a modal window that guides them towards upgrading their package or exploring alternative solutions. This will aid in enhancing the benefits of this community.
  • For integrations such as automation, real-time tasks can be added to the queue. For instance, automatic responses, API interactions, or other use cases can trigger notifications to users via the website, especially when error messages are returned. As a marketing strategy, users can be informed via email when such events occur, and alternative packages or solutions can be offered through email.

I am contemplating this audibly; more experienced colleagues may have better suggestions regarding its feasibility.

Thank you for your time, wishing you success <3