Respond to the right comments while your Instagram Live is still running. Match a keyword or every comment, reply publicly, move eligible conversations into DM, flag the ones that need a person and review every outcome in one event log.
Repli listens for Meta’s live_comments webhook event while the broadcast is running. Replay comments remain normal post comments and follow your regular comment rules.
Answer the viewer below their Live comment with a fixed message or a short AI response. Public replies are limited to 300 characters in the builder.
Use Meta’s private-reply route to send the configured DM when the comment is eligible. The system never promises a DM when Meta does not allow delivery.
Leave the fixed public reply empty and let AI draft a concise answer from the viewer’s comment and the instructions you provide. AI usage draws from your credit balance.
Set the maximum number of processed replies per minute for each rule. Additional matched comments are logged as rate limited and flagged for review.
Attach labels such as live-lead, launch or webinar to matched events so the team can understand why the comment mattered later.
Mark matched comments as needing human follow-up. Team members can acknowledge each event after they have handled it.
If Meta does not accept a public or private reply on the first attempt, Repli places the action into the existing job queue for a limited retry with backoff.
Viewers type LINK or PRICE as products appear. Repli acknowledges the comment publicly and attempts the eligible private reply with the relevant next step.
The host asks viewers to type NOTES. Matching comments receive the resource route and are tagged so the team can review interested viewers after the session.
AI can draft short public answers from approved instructions while comments that need judgement are flagged for a human instead of disappearing in the stream.
Words such as VISIT or DETAILS identify higher-intent viewers. The rule replies in public, offers an eligible DM handoff and marks the event for follow-up.
Instagram Live comments move faster than an ordinary post thread. A host is speaking, viewers are reacting, questions are stacking up and a useful buying signal can disappear above the fold in seconds. Repli’s Instagram Live comment automation listens for the official Live comment event from Meta, matches the text against rules for the connected account and carries out only the actions selected by the account owner.
The feature is not a fake comment generator, an engagement bot or a tool for posting messages as viewers. It responds to comments that real people write during the broadcast. A rule may post a public reply, attempt an eligible private-reply DM, generate a short public answer with AI, apply tags to the event or flag it for a person. Each result is recorded, including skipped and rate-limited comments.
The timing boundary is important. Meta delivers the live_comments event while the broadcast is active. After the Live ends, a new comment on the saved replay is an ordinary media comment. Repli leaves that to the normal Instagram comment automation system. This prevents one comment from wandering through two different rule sets and gives teams a clear place to troubleshoot each kind of interaction.
A strong keyword is short, easy to pronounce and directly tied to the host’s next sentence. “Type GUIDE and we’ll send the worksheet” is clearer than asking viewers to copy a long phrase. Repli accepts several comma-separated keywords, so reasonable variants can share a rule. Matching is case-insensitive in the current editor, which means GUIDE and guide behave the same way.
Use separate rules when the intended actions differ. PRICE might send a price page, BOOK might start a booking conversation and QUESTION might only flag the comment for a team member. A broadcast-specific rule can take priority over an account-wide rule, and an exact keyword rule takes priority over a wildcard. That order keeps a general catch-all from swallowing a more useful campaign response.
Reply-to-every-comment is available, but it deserves restraint. During a busy broadcast, answering every emoji and greeting can make the thread unreadable and consume the account’s message allowance. Use it for tightly moderated sessions where every comment needs an acknowledgement, set a sensible per-minute cap and watch the event log during the first run.
A public reply reassures the viewer that the host saw the request and keeps the context visible to everyone watching. It works well for confirmations such as “Sent the details” or concise answers that benefit the room. Repli limits the configured public reply to 300 characters. That encourages a response suitable for a moving Live thread rather than a paragraph that competes with the broadcast.
The private handoff is for details that belong in a one-to-one conversation: a link, a booking step, a product answer or the start of qualification. Repli uses Meta’s private-reply mechanism for an eligible comment. It does not bypass Meta’s delivery rules, and it cannot guarantee that every comment can receive a DM. A private reply is tied to the comment and is not permission for unlimited unsolicited messaging.
Many useful rules do both. The public line says what happened; the DM provides the next step. If the DM fails on the first call, the result is recorded and the action can enter Repli’s existing retry queue. If the comment is not eligible or Meta rejects delivery, the event remains visible instead of being counted as a successful conversation.
AI is optional. If the host asks viewers to type one keyword for one asset, a prepared response is faster to review and cheaper to run. AI becomes useful in a question-and-answer format where the viewer’s words matter and a single template would be evasive. In that case, leave the fixed public reply empty, enable AI and give it narrow instructions for the broadcast.
Good instructions include the topic, facts the assistant may state, answers it must avoid and the desired length. For example: answer only questions about the demonstrated product, use one sentence, never guess stock, and flag refund or safety questions for a person. Repli passes the viewer’s comment and username context into that instruction and truncates the public result to the allowed length.
AI answers use AI credits. They are not unlimited merely because the underlying comment rule is active. For facts that change quickly—availability, price, delivery time or legal terms—use a fixed link or route the viewer to a human unless the AI has an accurate source. A polished wrong answer is still wrong, and a Live audience will see it immediately.
A Live can jump from ten viewers to hundreds with little warning. Sending a reply for every comment in one burst is not good audience behaviour and may collide with platform limits. Each Repli Live rule has a configurable maximum replies per minute. The default is twenty in the editor, and the owner can choose a value from zero to 120. Zero removes this rule-level ceiling, but Meta’s limits and Repli’s account-level controls still apply.
When the rolling one-minute count reaches the rule’s ceiling, Repli does not pretend the next comment was handled. It records the event as rate limited, includes the reason and marks it for attention. This gives the team a factual list of missed opportunities after the broadcast. It also makes it possible to tune the cap based on the session instead of guessing whether the automation was too aggressive.
Outbound calls also pass through Repli’s shared Meta request controls, which respect provider rate-limit responses and apply backoff. Failed public replies and private replies can be queued for a limited retry. The queue does not create duplicate sends: the Live event table has a unique key for the connected account and comment, and the handler claims that event before it performs an outbound action.
The Live Comments area refreshes its event list regularly while the page is open. Each row shows the viewer, comment, time, matched keyword and processing status. Action chips show whether the public reply, DM handoff, AI response, tag or attention flag was sent, queued, skipped or failed. A failure reason stays attached to the event rather than being hidden behind a total counter.
Tags make the log easier to scan after a launch. A product team might use product-a, pricing and hot-lead. A webinar team might use attendee-question and follow-up. These are labels stored on the event; they are not presented as an automatic sync to an external CRM. Attention flags give the team a simple queue, and “Mark as handled” records when someone has reviewed an item.
The log is scoped to the selected Instagram account and its owner. Rules are queried by both user and connected account in the dashboard, while the webhook resolves the account Meta addressed before it looks for a rule. That boundary prevents a Live comment received for one business from triggering another business’s response.
Before announcing the keyword, open the Live Comments area and use the readiness checklist. Confirm an Instagram account is connected, at least one Live rule is active and the rule has an action. A rule that has no public reply, no DM text and no AI response can log a match but cannot send anything. The checklist calls this out before the audience arrives.
Start a private or low-audience test Live and comment from a second account. Do not test by commenting as the broadcaster because Repli deliberately skips the broadcaster’s own comments. Use the exact keyword first, then a different case, then a comment that should not match. If reply-to-all is enabled, verify the non-keyword comment follows the catch-all path.
Watch the event log. An event proves Meta delivered the webhook. A missing event points to the account connection, subscription or trigger. A logged event with a skipped reason points to the message allowance, phone requirement or rule cap. A queued action means the first Meta request failed and Repli scheduled a retry. This sequence is faster and safer than editing the rule repeatedly without knowing which step failed.
For a product demonstration, the host can say: “Type SIZE and we’ll send the chart.” The public reply confirms that the message is on its way, while the private reply contains the chart link. If a viewer instead asks a nuanced product question, a separate QUESTION rule can flag it for the host or allow a tightly instructed AI response.
For an educational Live, NOTES can deliver the promised resource and tag the event as workshop-notes. The first DM should provide exactly what was promised before asking for anything else. If lead collection follows, explain why each detail is useful and keep the number of questions proportionate to the value of the resource.
For a property tour, VISIT can open an eligible private conversation about a viewing. The system should not invent availability. The DM can ask for a preferred day or link to the correct booking page. The event can be flagged for attention so a person checks the conversation after the broadcast. This is what automation is good at: catching intent and organising the next step, not pretending every decision can be made by a bot.
The viewer experience may look similar—someone comments and receives a response—but the event source and operating controls are different. A Live comment is attached to a broadcast that is currently running. The dedicated Live workspace adds a readiness checklist, per-minute rule cap, attention workflow and event log designed for a short, high-volume session.
A regular post or Reel can keep collecting comments for months. Those rules are better suited to persistent keywords, comment-to-DM delivery and post-specific campaigns. Comments on a saved Live replay follow that normal path. Keeping the pages and product areas separate helps users choose the correct automation without creating two rules that they expect to fire on the same event.
Story interactions are different again because they arrive through Instagram messaging and lead directly into a private conversation. If the campaign starts with a Story reply or Story mention, use the Instagram Story automation page. If it starts with an ordinary DM keyword, the broader Instagram DM automation page is the right guide.
It is a rule that reacts to a real comment received while your Instagram Live is running. Repli can match a keyword or every comment, post a public reply, attempt an eligible private-reply DM, apply tags and record the result.
Yes. Choose one or more keywords or switch on reply-to-every-comment, then configure a fixed public reply or optional AI instructions. The automation applies only while Meta identifies the comment as a Live comment.
It can trigger an eligible private reply through Meta’s supported endpoint. Repli records whether the DM was sent, queued or failed and does not claim delivery when Meta does not allow it.
No. Comments posted after the broadcast becomes a replay are normal media comments. Handle those with a regular Instagram comment or comment-to-DM automation.
The system claims each comment with a unique connected-account and comment-ID record before sending. A repeated webhook delivery reaches the same key and stops before another action is sent.
The matched comment is logged as rate limited with the configured per-minute reason and is flagged for attention. Repli does not label it as successfully processed.
Yes. AI can generate the public response when the rule has AI enabled and no fixed public reply. Keep the instructions narrow and factual. AI responses use the account’s AI credit balance.
You can attach your own labels to each matched event and flag it for human attention. Those labels appear in the Repli event log; they are not described as an automatic external CRM sync.
Repli intentionally skips comments posted by the broadcasting account. A second account reproduces the viewer path and confirms that Meta delivered the event, the keyword matched and the selected actions ran.
The dashboard treats Live Comments as a Pro feature. Use the pricing page for the current plan comparison; access is determined by Repli’s shared entitlement catalogue.
Pricing · All features · How it works · Blog