Hi,
I would like some advice for how to tackle a feature that I would like to use in my project. The platform allows users to collaborate, and they can communicate by posing questions in a forum that we built into the platform, or individually or group messaging one another via the Zerocode Air Chat and Messaging Pro plugin that I have installed.
I would like to have an LLM (likely either ChatGPT or Claude) be a “user” of the platform such that users can:
- Message them directly in Air Chat like they would any other user and get a response, and eventually
- Allow the LLM to monitor discussions in the forum and post relevant responses in threads
I don’t even know where to start wrapping my mind around this. Can anyone suggest a way to go about tackling this, and resources or links that I can consult to help me with implementing this functionality?
Don’t use any plugin for this. I mean their plugin description is strange enough 'There is no more need for complex Bubble functionality involving repeating groups, inputs, and buttons. ’ - complex Bubble functionality? These are the basic features!
Have a dummy user account (ai@yourdomain.com). When a user creates a conversation with this account, only when recipient’s email = ai@yourdomain.com → run a workflow in the backend to generate an AI response
You could have a ‘cron’ AI which checks the forum every 5 minutes to see if there’s anything it wants to respond to, but I wouldn’t recommend that. Instead, when a reply is posted, using a DB trigger, make an AI call that first determines if an AI response is necessary based on the response and the thread it’s in. If no, don’t continue. If yes, generate a response and post it from the ai@yourdomain.com user (or just have an isAI (yes/no) on the Message data type and show a different name + avatar if yes).
2 Likes
