Does it work with a personal Instagram account?

No, and that is Meta's rule rather than ours. You need a professional account — Business or Creator, either is fine — connected to a Facebook Page. It is the single most common reason a first setup stalls before anything has been built.

Can it reply to comments on other people's posts?

No. The Instagram Graph API only lets an app act on the account that authorised it, so we can answer comments on your posts and messages sent to you. Any tool offering to comment on strangers' posts at scale is not using the official API.

Does it post, schedule or write captions?

None of the three. We answer people; we do not publish. There is no scheduler, no caption or hashtag generator, and no content calendar. The old version of this product had screens for some of that with nothing behind them, and they were deleted rather than rebuilt.

What happens if the AI is slow or unsure?

Your own template goes instead. A guarded generation that cannot be grounded in something you actually wrote falls back to the written reply rather than improvising, so something correct goes out or nothing does — never something wrong with your name on it.

Features — everything it does, and what it refuses

Answering

Comment automations. A trigger — one post or a set of them, plus keywords matched any-of or all-of — a public reply under the comment, and a private reply in the inbox. Negative keywords stop a rule that would otherwise fire. Matching ignores case and punctuation.

DM automations. The same shape, on messages sent to you, inside the window Instagram allows.

Comment to DM. The public reply is a receipt; the private message carries the answer. One rule handles both, because they are two halves of one action rather than two automations to keep in sync.

Try It. Type a message yourself in the builder and see whether the rule would fire, on which keyword, and exactly what would go out. Nothing is sent and nothing is recorded. Test on myself does the real thing to your own account before anyone else sees it.

Deciding

Ten checks run before a word is drafted, in a fixed order, and any one of them can stop the send: kill switch, connection health, human takeover, the messaging window, dedupe, per-person cooldown, quiet hours, rate budget, credits, content safety.

The ten checks every reply passes through, in order: kill switch, connection, takeover, window, dedupe, cooldown, quiet hours, rate budget, credits, content safety. — One gate, one order, one place that knows why something did not send. Every refusal is written down with the sentence explaining it.

Takeover is worth calling out. The moment you reply to a thread by hand, the automation stands down in it. It is a gate check rather than a setting, which is what makes "the tool never talks over me" a guarantee instead of an intention.

Sounding like you

Voice DNA is built from replies you actually wrote, not from a tone picker. Voice rules sit beside it for things you want stated rather than inferred — words you never use, the sign-off you always use.

Knowledge holds the facts a reply should be grounded in: your prices, your timings, your policies. A generated reply that cannot be grounded falls back to your written template. The guide on AI replies walks through setting this up.

Contact memory means a returning person is not asked something they already told you.

Seeing what happened

Activity lists every run — what fired, what went out, and what was refused with the reason. The blocked view is the same list filtered to refusals, each row naming the check that stopped it and what to do about it.

The inbox is the conversation view, with takeover and the window clock on each thread. Leads captures whoever triggered an automation with their handle, their exact words, the post it came from and the time, and exports as CSV.

Your name on your links

Branded links puts your own domain on every link an automation sends: yourbrand.com/l/… instead of postengage.ai/l/…, with HTTPS issued and renewed for you. It comes with link insights: taps per day and hour, device, the app a link opened in, country, and which link and rule did best. Instagram's own link-preview fetches are filtered out of every count, on every plan. Included in Studio, or $7 a month as an add-on; the setup guide takes two DNS records.

Running it with other people

Named seats with roles, on the plans that include them. Administrative actions are written to an append-only record. Impersonation by our own staff requires a reason and leaves a trail — the security page sets out how.

Your data

Export everything held about a person as a file. Erasure walks a fixed list of tables in a fixed order, and it is the same list that drives the export, so what you can download and what gets deleted cannot drift apart. Details on the privacy page.

What it does not do

The important half.

Not built

Post scheduling. Caption and hashtag generation. A content calendar. Reel performance prediction. Follower-growth features. An analytics product.

Why

We answer people; we do not publish. Three of these existed in the old build as screens with no working backend, showing numbers that were never measurements.

Also not built

WhatsApp, Messenger, Telegram. A visual flow-builder canvas. Broadcasts. Multi-step nurture sequences. One-click CRM integrations.

Why

One channel, done properly, is the whole bet. The canvas and the broadcast are how a messaging tool turns into something that gets accounts restricted. Leads leave as CSV.

If your setup depends on any of those, another tool is the better buy and the comparison pages name which one, on each page, specifically.

The platform limits you inherit

Not ours, and no tool can remove them.

A professional Instagram account connected to a Facebook Page is required. A comment can be replied to for seven days. A DM can be sent only within twenty-four hours of that person's last message, and only they can restart that clock. We can act on your account and nobody else's.

What the two windows mean in practice is worth reading before you build anything, because almost every "why didn't it send" question turns out to be one of them.

Where to start

The comment automation guide builds a working rule in about ten minutes, and the pricing page explains what any of it costs — which for a template-only setup is nothing.

Everything you need to sell in the comments.

Reply in public. Send the link in private.

Written the way you actually talk.

Ten checks before anything is sent.

Every commenter becomes a lead.

Every conversation, one place.

See every decision it made.