Automations
Keyword-triggered Instagram replies. Someone comments "GUIDE" on your post, they get a DM with the link.
Studio, then Automations. Requires a connected Instagram account.
The rule that makes this safe
Automations send YOUR text, never AI text
The automation engine never generates or rewrites anything. It sends exactly the reply you wrote.
Your approval is the moment you activate the automation, and that is audited: who activated it and when. New automations are always created inactive.
The engine has no path to any AI service at all.
Triggers
| Trigger | Fires on |
|---|---|
comment_keyword | A comment on your post |
dm_keyword | A direct message |
story_reply | A reply to your story |
story_mention | A story mentioning you |
dm_default | Any DM that matched nothing else |
Matching
Keywords, up to 20, matched as contains, exact or starts_with. Up to 20 exclusion keywords.
A comment automation with no keywords runs on any comment. A DM keyword automation requires at least one.
Scope to specific posts by picking them, up to 50, or leave it open to all posts.
First match wins, then evaluation stops. Post-scoped comment automations are tried before all-posts ones. Plain DMs try the keyword automations first and fall through to dm_default afterwards.
Actions
Send a DM, up to 1,000 characters, with up to 3 link buttons.
Reply publicly to the comment, up to 5 variations of 300 characters each, one picked at random so your comment section is not full of identical replies.
Apply tags to the contact.
Everything also lands in your Inbox, so you see the conversation.
Opt-in flow
Instead of DMing immediately, send a button. The DM is delivered when they tap it.
This is worth using: an unsolicited DM is more likely to be marked as spam than one the person asked for.
You can also gate on following you, with your own message for people who do not.
Guards
Several, and they matter:
| Guard | What it prevents |
|---|---|
| Redelivery check | The same person triggering twice on the same comment |
| One private reply per comment | Meta's hard limit, across all your automations |
once_per_contact | Repeat sends to the same person |
| Cooldown | Rapid repeat triggering. Required for dm_default, 30 minutes to 7 days |
| Follower gate | Sending to people who do not follow you |
| Require and skip tags | Running for the wrong segment |
| Self-comment guard | Your own comments triggering your own automations |
Activating
Automations are created inactive. Toggle to activate, which is the approval moment and is recorded.
Your plan caps how many can be active at once. Hitting it names the limit and asks you to pause one or upgrade.
When sending is off
There is one switch, on the server, for everything Postlyra sends to a social platform: posts, comments, reposts, DMs and automation replies alike. It is off outside production. While it is off, every match is recorded as logged, not sent: the run row shows exactly what would have gone out, with nobody actually messaged.
This is not something you configure per brand or per automation; it is a property of the environment the server runs in. On production it is on, so an activated automation sends for real.
Activity
The second tab shows every run: sent, logged (would have sent, but sending is off), skipped (with the reason) or failed.
The skip reasons are the useful part. They tell you which guard stopped a run, which is nearly always the answer to "why did this not fire".
Syncing posts
To scope an automation to specific posts, sync your existing Instagram posts into the picker. Throttled to once every 5 minutes per brand.