01
Master on/off
A single switch turns AI replies on or off across every connected Instagram account.
AI disclosure
Honest about the AI behind the replies.
ReplyMagic generates messages that go out from your account. You should know exactly how that works.
How it works
What we do, in plain English.
"The model only sees the message we're answering, the post it was made on, and the context you've configured."
ReplyMagic uses general-purpose large language models from established commercial AI providers — such as OpenAI, Anthropic, and Google (Gemini) — to draft replies to Instagram comments. We don't pin our copy to a specific model; we evaluate options over time and route to whichever performs best for the task.
To keep replies on-topic, the agent can also analyze the media on a post — the image, carousel, or video — using a commercial AI provider, so it understands what the post is actually about before answering.
What the model does not see: other customers' data, scraped public profiles, or any private conversations outside the comment we're answering and the post it was made on.
Training
Your context is yours.
We use the context you provide only to generate replies for your account. We do not sell your data, and we do not use it to train our own models. Your content is sent to AI providers through their business APIs and is processed under those providers' API terms and privacy policies.
Controls
Controls that actually exist in the app.
Everything below ships today on the AI settings page.
02
Voice persona
Pick a preset (brand, founder, expert, creator, mascot, pet, custom) so the assistant sounds like you.
03
Tone & formality
Describe the voice in your own words and pick a 1–5 formality level.
04
Reply length
Choose short, medium, or long for outgoing replies.
05
Emoji & hashtags
Toggle either off if they don't fit your brand.
06
Signature moves
Short patterns the AI should use when relevant ("invite to DM for sizing", etc.).
07
Hard don'ts
Patterns it must never use ("don't quote prices publicly", etc.).
08
Goals
What each reply should be moving toward — bookings, DMs, link clicks, etc.
09
Exclusion phrases
Comments containing any listed phrase are skipped by the AI entirely.
Limits
What AI can and cannot guarantee.
Large language models can make mistakes — wrong facts, off-tone phrasing, or replies that don't fit a particular comment. We work to reduce this with the controls above, and you can edit your AI settings or turn AI replies off at any time.
You are responsible for the replies sent from your connected accounts. We strongly recommend reviewing the activity feed regularly, especially when you first connect an account or make significant changes to your context or rules.
Questions about how this works on your account? Write [email protected].