S:01
Drafted
The AI has read the comment, pulled context, and generated a reply. It's waiting for your call.
Awaiting review
Feature · Approval workflow
Stay in control of every reply. Or none. Your call.
ReplyMagic is hybrid by default. Trusted categories auto-send. Edge cases queue. Sensitive topics always route to you. The activity log keeps every state explicit so nothing is invisible.
Reply lifecycle
Six states. Each visible.
Every comment that hits ReplyMagic moves through one of these states. The activity log shows exactly which state every comment is in, with timestamps and the reasons behind every routing decision.
S:02
Approved
You hit approve. The reply moves to the send queue and goes out within seconds, paced naturally.
Ready to send
S:03
Edited
You tweaked the draft before approving. Edits are saved as future voice samples — the AI learns from your touch.
Voice learning
S:04
Rejected
Not the right reply. ReplyMagic learns to avoid that pattern, and the comment is marked seen so it stops re-surfacing.
Pattern signal
S:05
Sent
Reply is live on Instagram. Logged with the post, comment, response time, and channel (comment vs DM).
Audit trail
S:06
Skipped
Spam-gated, policy-flagged, or out-of-scope. Skipped comments show up in the activity log so nothing is invisible.
Transparent
Guardrails
Set the lines. The AI respects them.
Your guardrails sit at the top of every prompt. They override context, voice, and language. If a comment lands inside a guardrail, it routes to you — never to a guess.
G:01
Hard don'ts
Phrases or topics ReplyMagic never says — pricing speculation, medical claims, refund promises. Hard-coded in your settings.
Phrase blocklist
G:02
Refer-to-you topics
Topics that always route to your queue: refunds, custom orders, brand collabs, anything sensitive. The AI never auto-answers these.
Manual route
G:03
Signature moves
Phrases or rituals you always use — 'sending a DM with the link 💌' — added to the voice profile so they appear naturally.
Brand consistency
A typical setup
What most creators actually run.
Week 1: review-mode. Every reply queues. You approve or edit each one. The AI absorbs your edits as voice samples. By Friday, drafts are landing in your voice 9/10 times.
Week 2: hybrid mode. Auto-send turns on for the safe categories — price questions, sizing asks, link requests. Pricing edge cases, refund mentions, and DM-only follow-ups still queue.
Week 3 onwards. You check the queue once a day. ReplyMagic handles the long tail in your sleep. The four real conversations a week still come to you with full context.
FAQ
Common questions about approval.
Do I have to approve every reply? +
No. Most accounts run a hybrid: trusted categories (price asks, sizing, link requests) auto-send, and edge cases queue for review. You set the rules.
Is there a mandatory review window? +
No. We recommend running review-mode for the first week so you can check the voice match — but it's optional. Some accounts auto-send from day one. Some keep manual approval forever for high-stakes brands. Your call.
Can I edit a draft before approving? +
Yes. Every queued reply is editable inline. Edits are stored as voice samples — over time, the AI absorbs your tweaks and the queue shrinks.
What about replies on phone? +
The activity log and approval actions work fully on mobile. Most creators triage from phone in spare moments — five seconds per reply, queue empty by the end of a coffee.
Is there an audit trail? +
Yes. Every comment carries a status (drafted / approved / edited / sent / skipped) and a timestamp. Useful for agency QA and for understanding why a reply went out a particular way.