--- title: "DM Setter for Instagram: Sort, Answer and Book DMs" slug: dm-setter-for-instagram author: "Leonardo Maldonado" category: "AI Setter Operations" articleType: how-to-guide tags: ["Instagram DM triage","automated Instagram front desk"] publishedAt: 2026-10-26T09:00:00.000Z updatedAt: 2026-10-26T09:00:00.000Z canonical: https://setluca.com/blog/dm-setter-for-instagram ---# DM Setter for Instagram: Sort, Answer and Book DMs > A DM setter for Instagram does the front-desk job on your DMs: it greets whoever writes in, works out what they want, and answers the questions you answer the same way every time. It takes a proper message when only you will do, and puts the booking through to your calendar when they're ready. The coach stays the person the call gets transferred to. The setter just makes sure nobody is left waiting. ## Key takeaways - A DM setter's first job is sorting, not selling. Most of what lands in a coaching inbox is never a sales conversation. A front desk that treats every message as a lead is worse than no front desk. - Meta gives you 24 hours to respond to an Instagram message, and a Human Agent tag that stretches the door open to seven days. - Messages from people who don't follow you land in the Requests folder, and Meta's API stops returning them after 30 days of inactivity. - Say it's automated. The EU's transparency rules for AI systems that talk to people took effect in August 2026. Meta's own developer policies tell you not to confuse or mislead anyone. - Every message the setter can't finish becomes a message slip for you, with the thread attached. That queue is the product, not a fallback. --- ## What is a DM setter for Instagram? A DM setter for Instagram is software on the front desk of your DM inbox. It does the five things a front desk does: greet, triage, answer the known questions, take a message, transfer the call. It replies to people who message you first, and never dials out. Think of it as a front desk. Nobody hires a front desk to close deals. You hire one so the inbox stops pinging you at 11pm. And so the person who wants to buy something reaches a human while they still care. The software's mechanics live in the walkthrough of [how AI appointment setters work](/blog/how-ai-appointment-setters-work). This page is about the desk. ## What are the five jobs a front desk actually does? Five, in this order, and only one is about selling. | Front-desk job | On a phone | In your Instagram DMs | | --- | --- | --- | | Greet | "Good morning, Ridgeway Physio." | A first reply naming the business | | Triage | "New patient or existing?" | New enquiry, client, pitch, fan mail, media | | Answer the known question | Hours, parking, address | Price band, program length, what's included | | Take a message | Name, number, reason for calling | Goal, timeline, budget signal, timezone | | Transfer the call | "Putting you through now." | A booked slot, or a slip in your queue | The greeting is the cheap part, and where most tools stop. The value sits in triage. A desk that greets everyone warmly, then drops 400 unsorted messages on you, has moved the work rather than done it. ## Step 1: Write the greeting before you write anything else Write one greeting per surface, and make it sound like your business rather than a support ticket. Someone replying to your story is in a different mood from someone who commented "price?" on a reel, and one flat hello for both reads as machinery. Three things belong in a greeting: who they've reached, that the message landed, and what happens next with a rough time. No qualifying question. It lands like an interrogation before hello. > "Hey. This is the front desk for Coach Priya's inbox, answering on her behalf. Got your message. I can sort out most things here, and anything that needs Priya gets to her the same day. What are you working on right now?" Instagram's own instant replies handle this much without any external tool, and their setup is covered in the guide to Instagram auto reply. ## Step 2: Build the triage buckets before you write any replies Sort first, write second. Almost every coaching DM falls into one of six trays, and the desk's accuracy comes from how well you define them, not how well the AI writes. New enquiry. Current client. Pitch or spam. Fan mail. Media or collab request. Something urgent and personal. Give each a test software can apply, like "is this handle in the client list" rather than "sounds like a client." Only the first tray is a sales conversation. That split is the point of the front-desk frame. It's what separates a well-run DM setter from a setup where every thread is a prospect until proven otherwise; [what a DM setter is](/blog/what-is-an-ai-dm-setter) covers the basics. ## What happens to the DMs that are not leads? They get handled, and none of them get sold to. This is the half of the inbox nobody writes about, and it is most of the inbox. Current clients get answered or routed, never pitched. If your support lives elsewhere, the desk's job is one line pointing them there. Pitches and agency spam get filed with no reply, because replying tells the sender the address is live. Fan mail gets a short acknowledgement, media requests a flag with the ask summarised. The urgent tray needs a hard rule. If a message mentions self-harm, a medical emergency, a bereavement, or a dispute about money already paid, the desk stops, sends nothing beyond an acknowledgement, and pages you. Software has no business improvising there. Holding every channel's trays in one place is easier with a [unified inbox for coaches](/blog/unified-inbox-for-coaches) than across four apps. ## Step 3: Answer only the questions with a stable answer A good front desk quotes opening hours, not treatment plans. The test is whether your answer would change based on who's asking. If it would, the desk takes a message instead. Safe: what the program is, how long it runs, roughly what it costs, what's included, whether you work in their timezone. Not safe: whether their injury is something you'd take on, payment plans, refunds, anything medical, and any price you'd flex on. Keep the list short. Eight to twelve stable answers cover a coaching inbox, and each one past that is another chance for the desk to be confidently wrong. ## Step 4: Take a message properly A message slip beats a transcript. When the desk can't finish a conversation, what reaches you is a five-field slip you can read in ten seconds. The fields are who they are, what they want, what they've tried, when they're free, and what the desk already told them. The last field, what the desk already told them, is the one people forget, and the one that stops you contradicting your own front desk. Which questions earn a place on the way to the slip is covered in [AI lead qualification in DMs](/blog/how-ai-qualifies-leads-in-dms). The message slip is also why we treat the queue of handoffs as the product rather than a safety net. A desk that acts on everything is a liability. One that sorts everything and hands you the exceptions is staff. ## Step 5: Transfer the call The transfer is either a booked slot or a named handoff, never a link dumped in the thread. If the lead is ready and meets your criteria, the desk offers real open times and writes the event. The guide to [how AI books calls from your DMs](/blog/ai-that-books-calls-automatically) covers that mechanism. If the lead misses your criteria, the desk says a human is picking this up and gives a window. Say the handoff out loud. "I'm passing you to Priya, she'll be in here this afternoon" resets expectations and stops the lead re-explaining everything when you arrive. ## How long does Instagram let the front desk stay on the line? Not indefinitely, which is the real reason a front desk beats a coach checking DMs twice a day. Meta's Messenger Platform policy gives businesses up to 24 hours to respond, and promotional content is allowed inside that window. Miss it and the standard door shuts. Three other clocks matter: - **Under the same Meta policy, the Human Agent tag** stretches the window to seven days, so a human can take over a thread the automation started. - **Private replies**, under Meta's private replies documentation, allow exactly one message to someone who commented on a post or reel, within seven days of the comment, and continue only if they answer. The [comment-to-DM automation](/blog/comment-to-dm-automation) playbook covers what fits in that one message. - **The Requests folder** holds every new conversation from a non-follower. Meta's documentation notes that requests with no activity for 30 days stop being returned by the API. Lollipop chart comparing four Instagram messaging windows in hours: standard reply window 24 hours, Human Agent tag 168 hours, private reply window 168 hours, and Requests folder inactivity cutoff 720 hours ### What the throughput ceilings allow Throughput has ceilings too. Meta's Instagram Platform documentation lists two calls per second per account on the Conversations API and 750 private replies an hour. A desk paced like a person never gets near those. A blaster does. The speed argument is settled in the piece on [DM response time and speed to lead](/blog/dm-response-time-speed-to-lead). Coverage is this page's claim, and it is a claim about being reachable rather than about outcomes. ## Does the setter have to say it isn't you? Yes, and you should want it to. Writing in your first person, as you, turns a helpful front desk into a betrayal the moment someone works it out. Meta's developer policies carry no explicit "label your bot" clause today. What they carry is broader: policy 2.4 says plainly, "Don't confuse, deceive, defraud, mislead, spam or surprise anyone." A DM presenting automation as the coach sits inside that, no bot-specific rule needed. ### The EU now requires the disclosure outright The law has gone further than Meta has. Article 50(1) of the EU AI Act covers any AI system built to interact directly with people. Providers must "inform natural persons that they are interacting with an AI system, except where this is obvious from the circumstances and the context of use." The European Commission confirms those transparency rules took effect in August 2026. If EU leads write to you, this applies to your desk. ### How to disclose it, and how to recover if you didn't Disclosure costs one clause in the greeting and nothing after, and you never deny it when asked. If someone types "am I talking to a bot," the answer is yes, plus an offer to fetch the coach. Sounding human while staying honest about not being the coach is its own craft. The guide to [DMs that don't sound like a bot](/blog/make-ai-dms-sound-like-you) covers it, and our read of [Meta's business messaging automation policy](/blog/what-meta-allows-for-dm-automation) covers Meta's wider position. When disclosure goes wrong, recovery is short: the coach picks up the thread by name, says a person is here now, and answers the question. People forgive a front desk. They don't forgive being lied to about who they were talking to. ## What does one month of Priya's front desk look like? Illustrative, with round numbers from a strength coach's inbox rather than a real account. Priya gets 380 inbound Instagram threads in a month. Sorted into trays, that month holds: - 141 story replies - 62 questions with stable answers - 47 messages from current clients - 34 pitches and agency spam - 29 private replies off comments on two reels - 67 genuine new enquiries Of the 141 story replies, 88 are one word or an emoji, so they get an acknowledgement and stop. The other 53 get a real reply, and 11 become enquiries folded into the 67. The 47 client messages go to her client channel and never see an offer. The 34 pitches get nothing, and her list of ten stable answers covers the 62 questions. Nine of those people then ask something it can't cover, so those become slips. ### Where the 67 real enquiries end up That leaves 67 enquiries. Of those, 21 book directly and 18 get flagged because the answer depends on Priya. The other 28 go quiet and enter the rescue cadence, which stops rather than running forever. The number worth staring at is 313. Those messages were never a sales conversation, and the front desk handled them so Priya's evening didn't have to. ## What edge cases break a naive front desk? **The 2am reply from another continent.** A lead in Singapore answers while you sleep. The desk should reply at their hour, and never say "first thing tomorrow" without knowing whose tomorrow that is. Store the timezone the moment you have it. **The current client who arrives as a stranger.** They message from a second account, or their handle changed. The check misses and the desk offers a discovery call to somebody who has paid you for eight months. Check thread history before routing, and bail out to a slip on any ambiguity. **The request that sat in the folder.** A non-follower messaged 31 days ago, you never opened Requests, and the thread is now invisible to your tooling. Sweep it on a schedule. A desk that watches only the main inbox ignores the side door. ## Why isn't my Instagram DM setter replying? | Symptom | Likely cause | Fix | | --- | --- | --- | | Story replies never get answered | It was a reaction, not a message, or the sender isn't a follower and landed in Requests | Handle reactions separately, sweep Requests daily | | A reply sent a day later fails to deliver | The standard 24-hour window closed | Use the Human Agent tag for up to seven days, or reopen off a fresh comment | | Comment-to-DM fires once, then nothing | Private replies allow one message per comment | Write it to invite a reply, since their answer reopens the 24-hour window | | Sends slow to a crawl during a launch | Two calls per second per account, 750 private replies an hour | Pace the desk at human speed and queue the overflow | | A client asking about billing got a sales pitch | No client check ran before routing | Check the client list first, and keep that tray silent on offers | ## What mistakes turn a front desk into a liability? **The Fake Front Desk.** Writing in the coach's first person with no disclosure. It works right until it doesn't, and the recovery costs more than the messages were worth. **The Universal Greeting.** One hello for every message, including the one from a client whose father just died. Trays exist so greetings can differ. **The Answer Machine.** Loading the stable-answer list with things that aren't stable, usually pricing exceptions and anything medical. The desk answers confidently and wrongly, and you find out a week later. **The Silent Filing Cabinet.** Routing non-leads into a folder nobody opens. Media requests and client questions rot there. A tray with no owner isn't a tray. **The Instant Escalation.** Every message forwarded to the coach with a notification. That's a forwarding service wearing a front desk's badge. ## Should you run a front desk, a full sales setup, or neither? Run nothing automated under about 20 DMs a week if you answer them the same day. Turn on a greeting and stable answers at 20 to 100 a week, when the same four questions keep arriving. Run the full front desk past roughly 100 a week, or as soon as client and lead messages are tangled together. Move to a complete sales setup when enquiries alone eat more time than delivery. Hire a human instead if your inbox is small but each conversation is worth five figures. Same answer if your work is regulated enough that a wrong sentence is a legal problem. That trade-off is argued out in the piece on [whether AI DM automation actually works](/blog/does-ai-dm-automation-actually-work). ## Where does a human still beat a DM setter? At recognising the regular. A good front desk knows the person calling every Tuesday is the client whose program just got harder, and shifts tone before a word is exchanged. Software matches on handles and history. It doesn't hear that somebody sounds different today. Software also loses on the exception it has never met. A human improvises safely because they know what the business would want; AI improvises confidently, which is a different thing. That gap is why the message slip exists. Creators running their own desks hit the same wall, as the guide to an [AI DM setter for creators](/blog/ai-dm-setter-for-creators) lays out. ## Getting the desk staffed A DM setter for Instagram earns its place on the messages that were never going to become clients. Clear those, and the ones that might become clients reach you while they still care. Write the greeting, define six trays, list ten stable answers, decide what a slip holds, and disclose in the first line. A weekend of setup, and your evenings stop belonging to the inbox. Luca runs that desk the way you write. Auto-send is on by default on every channel, and you can switch any channel to review. Instagram and Messenger are live in Luca. Telegram is live today, and WhatsApp is coming soon. The coach stays the closer. Plans are $99, $249 and $499 a month for 1,250, 4,000 and 8,500 replies, and every Luca plan comes with a 7-day free trial (card required). [See how Luca handles your inbox](/pricing). ## FAQ ### What is a DM setter for Instagram? It's software that answers your Instagram DMs the way a front desk answers a phone. It greets whoever writes in, sorts them into the right tray, and answers the questions that have stable answers. When the coach is needed it takes a proper message, and when a lead is ready it transfers the booking to your calendar. ### Is a DM setter for Instagram allowed under Meta's rules? Yes, when it replies to people who message you first and stays inside Meta's messaging windows. Meta's policy gives businesses 24 hours to respond, extendable to seven days with the Human Agent tag. Its developer policies also require you not to confuse or mislead anyone. ### Does the AI have to say it's not the coach? Yes. Article 50(1) of the EU AI Act requires people to be told they're interacting with an AI system. Those transparency rules took effect in August 2026. One clause in the greeting covers it, and you never deny it when somebody asks directly. ### What happens to messages that aren't sales leads? They get sorted rather than sold to. The desk routes current clients to your support channel, files pitches without a reply, and gives fan mail a short acknowledgement. Anything urgent or personal goes straight to you, with nothing sent beyond a note that a human is coming. ### How is front-desk sorting different from appointment setting? Front-desk sorting covers the whole inbox; appointment setting works only the sales conversations and qualifies them. A good DM setter does both, and the sorting comes first, because it is what makes the selling manageable. ### What should I never let it answer? Anything whose answer changes depending on who is asking. Refunds, payment plans, medical or injury questions, disputes about money already paid, and any price you'd negotiate. Those become message slips with the thread attached. ## Sources 1. [Meta for Developers. "Messenger Platform and Instagram Messaging API policy overview."](https://developers.facebook.com/docs/messenger-platform/policy/policy-overview/) 2. [Meta for Developers. "Private Replies: Instagram Platform."](https://developers.facebook.com/docs/messenger-platform/instagram/features/private-replies) 3. [Meta for Developers. "Instagram Messaging API: Instagram API with Instagram Login."](https://developers.facebook.com/docs/instagram-platform/instagram-api-with-instagram-login/messaging-api) 4. [Meta for Developers. "Instagram Platform overview: rate limits."](https://developers.facebook.com/docs/instagram-platform/overview) 5. [Meta for Developers. "Meta Platform Developer Policies."](https://developers.facebook.com/devpolicy/) 6. [Publications Office of the European Union. "Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 50." Retrieved October 17, 2026.](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689) 7. [European Commission. "AI Act: Regulatory framework for AI." Retrieved October 17, 2026.](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai) --- Published by SetLuca.