How to maintain ongoing conversations for mobile users

API
1

Hello! I have a live mobile app on the Apple App Store for tracking ToDo’s It’s like a ToDo / habit tracker. It also has a beta Chat feature which I enabled via OpenAI API (GPT 4.0).

How would I be able to do the following:

  • Allow each user to have an ongoing conversation. So when they come back to the app, day after day, the Chatbot remembers the conversation, who they are (based on logged in UID) what they’ve talked about etc.?
  • Put some sort of limit on the amount of use per user, so I don’t get some users that overspend on the API so my costs don’t go through the roof?
  • Be able to say to the users that the info they provide is not shared as training data with OpenAI?

Apologies if this is a basic question.

2

OpenAI said they do not train their models on inputs and outputs through their API.

1 Like
3
  • Allow each user to have an ongoing conversation. So when they come back to the app, day after day, the Chatbot remembers the conversation, who they are (based on logged in UID) what they’ve talked about etc.?

Either create your own memory solution, or use something like Zep, or use Assistants API (in beta, not very stable).

  • Put some sort of limit on the amount of use per user, so I don’t get some users that overspend on the API so my costs don’t go through the roof?

Save usage linked to the user id and check before each request if the limit has not been reached.

1 Like
4

You must manage your own users by having them only communicate through your own moderated API by their own account.

You must never put your API key in client software.

You must never allow client software to make direct contact to OpenAI.

The results of ignoring this advice will be for your API account balance to be emptied by leaked key abuse.