To auto-reply to Instagram and Facebook comments without sounding like a bot, connect your accounts to a tool that uses the official Meta API, create one rule per post or per keyword, use an AI reply driven by a short written instruction, add a cooldown, and cap how many replies one person can get. Then keep complaints out of the robot's hands.
TLDR: Fixed templates sound robotic and look like spam when 40 people get the same line. An AI reply with a one-sentence instruction, a 1 to 3 hour cooldown and a max of 3 replies per person reads human. Complaints go to a person, never to auto-reply.
What Instagram and Facebook actually allow
Both networks let business and creator accounts answer comments through their official API. A tool connected to your account can:
- Post a public reply under the comment.
- Send a private DM to the commenter (Meta calls this a private reply). On Instagram this only works within 7 days of the comment, and only once per comment.
Personal Instagram profiles can't be automated this way: switch to a professional account first (it's free).
Native options are thin. Meta Business Suite has keyword automations, built mostly around DMs, with no cooldown, no cap per person and no approval step for a team. That's why most brands and agencies use a dedicated tool.
Step 1: decide what each rule answers
Start with one rule, not ten. The most useful rules we see:
| Rule | Trigger | Reply |
|---|---|---|
| Thank every commenter | All comments on a post | AI public reply |
| Send the link by DM | Comments with "link" | Public "Sent you a DM" + DM with the link |
| Answer questions | Comments with "?" | AI draft that a person approves |
Two scope settings matter:
- Specific posts. A launch post and a giveaway need different answers. A rule tied to one post can say "the drop opens Friday at 6pm" without that line showing up under your recipe video.
- Specific profiles. If you run several brands or clients, a rule should only fire on the accounts it was written for.
Step 2: write an instruction, not a template
After the fifth identical "Thanks for your comment! 🙏" in a row, readers notice, and so do the networks.
An AI reply writes a fresh answer for each comment from an instruction you give it. Keep the instruction short and concrete:
- "Thank them warmly in one short sentence and mention what they said. Use their first name if you know it."
- "Answer their question in one or two short sentences. If you are not sure, invite them to send us a message."
What makes it sound human:
- Ask for one or two sentences. Long replies are the clearest bot tell.
- Mention what they said. "Glad the blue one won you over, Sara!" beats "Thank you for your support!"
- Give it facts. If people ask about shipping, put the real answer in the instruction ("We ship to the US and Canada, 3 to 5 days"). Without facts, the AI guesses.
- Give it a way out. "If you're not sure, invite them to DM us" stops it from inventing an answer.
Step 3: add a cooldown and a cap per person
Cooldown. A reply that lands 4 seconds after the comment feels automated. A delay of 1 to 3 hours looks like a person catching up on comments, and it gives your team a window to answer first. If someone on the team replies before the cooldown ends, the auto-reply should cancel itself.
Max replies per person. If your AI reply asks a question, the commenter answers, the rule fires again, and you get a bot talking to one person forever. Cap it at 3 replies to the same person per post. Everyone still gets an answer; nobody gets ten.
Step 4: answer the comments already waiting
Most rules only catch new comments. But a post that went live yesterday may already have 60 comments with no reply.
Run the rule on those too, with three precautions:
- Space the replies out, one every 30 to 90 seconds or so, not 60 in one second. A burst is exactly what spam filters look for.
- Use the AI reply, not a fixed template, so 60 replies aren't 60 copies of one line.
- Remember the 7-day limit: comments older than a week can get a public reply, but not a DM.
What never to auto-answer
Some comments need a human, every time:
- Complaints: "My order never arrived", "Still waiting on my refund".
- Angry or upset comments, even short ones.
- Anything legal, medical or about safety.
- Press, partnership or wholesale requests.
A cheerful AI line under "you charged me twice" is worse than no reply. For these, build a rule that tags the comment and alerts a person instead of replying, using keywords like refund, broken, never arrived, charged, scam. For question rules, choose draft then approve, so a person reads the answer before it goes out.
How to set this up in Mydrop
In Mydrop, auto-replies live in the social inbox under Auto-reply. A rule works on Instagram and Facebook comments (and the other networks you connect):
- Click New rule, or start from an example such as Thank every commenter or Send the link by DM.
- Pick the trigger: all comments or specific keywords, on all posts or specific posts, for the profiles you choose.
- Choose the reply: AI reply with your written instruction, or a fixed template, plus an optional DM.
- Pick the reply mode: Auto reply, Draft then auto-send, or Manual approval. Set the cooldown and the max replies to the same person.
- Switch it on. If comments are already waiting, the rule shows Run on N now and answers them one at a time, spaced out.
You can also open a post's comments and click Create auto-reply. Every reply shows in the rule's Results. More on keyword rules on the trigger words page.
Your next step
Pick your best-performing post from the last 7 days. Write one AI instruction for it (two sentences, one fact, one way out), set a 2-hour cooldown and a max of 3, and run it on the comments already waiting. Read the first ten replies yourself, adjust the instruction, then switch the same rule on for your next post.
Want the moderation side too (hiding spam, routing leads)? See how to automate comment moderation.














































Google review
Trustpilot review