If the process crashes or the container falls over, the conversation will be interrupted with no clear error message, nor will it be able to recover from the stored state.
[ ] Ensure we are storing the conversation state into our database at every state of the conversation
[ ] Ensure that if no response is received, we flag this to the user
[ ] Build a suitable retry flow which can elegantly recover
If the process crashes or the container falls over, the conversation will be interrupted with no clear error message, nor will it be able to recover from the stored state.