Has anyone implemented an agent to handle Instagram DMs through ManyChat, but it's having trouble with multiple messages. People write "hello," "how are you?" on three different lines, and the agent treats it as three different messages. Does anyone know how to deal with this? I tried using a message buffer, but the output of that architecture is "do nothing," but it still sends an output message, which causes the agent to respond anyway, and the same thing happens.

Hi, I was wondering if someone might have experienced the 200 error. To create a message buffer for IG DMs, I have a webhook connected to the IG DMs with Manychat. These message buffers generally wait a certain amount of time to accumulate messages through a wait process before sending a single response for all the accumulated messages. The problem is that these messages, which in theory remain in the queue and end in a "no operation" in these flows, still send a 200 or 500 error as a response, which causes the previous message to be repeated in IG. This automatic response renders any possible buffer in n8n useless. Has this happened to anyone else or have they been able to fix it?

I once heard from the Open AI people that bots hallucinate, and that's why it's dangerous to believe everything they say. At the agent level, it's essential to be able to control these types of problems, which can be very delicate in areas like healthcare. Do you believe hallucinations are inevitable, or are there simple models or solutions to achieve high accuracy?