Meta allows businesses to automate Instagram DMs, with conditions. Reply Mitra follows them for you, and most of them cannot be switched off, so your account stays in good standing.
Say the chat is automated
When a conversation is held by AI, people must be told. The AI agent's first reply always says it is an automated assistant for your account and how to reach a person. If someone asks whether they are talking to a bot, it answers honestly.
Offer a way to a person
Anyone talking to an automated chat must be able to reach a human. In the AI agent and AI FAQs, handover to a person is always on. Phrases like "talk to a person" hand the conversation to you, and you get a notification.
Do not talk over a person
When you reply to someone yourself from Instagram, the AI agent and AI FAQs stop answering that person, so they never talk over you.
Reply promptly
Automated replies should feel like a reply, not a delay. Reply Mitra sends replies within seconds, the AI agent within about half a minute, and if the AI cannot answer it tells the person a human will follow up instead of leaving them waiting.
Only message people who contacted you
Reply Mitra only messages people who commented, replied to a story, or sent you a DM, inside Instagram's messaging windows. It never sends messages to people who have not engaged with you.
Your own messages still need to follow Meta's policies: no misleading claims, and no spam.