The AI DM setter that turns Instagram, Facebook and WhatsApp DMs into booked calls
ReplySetter replies to every DM within seconds, any hour of the day. It answers the lead's question, qualifies them the way your best setter would, handles the "too expensive" and "let me think about it" moments, follows up when they go quiet, and books the call straight onto your calendar. Every message, field update and booking is on the record, so you can see exactly why it said what it said.
- Build and test free
- DM channels on Basic ($97/mo) via GoHighLevel
- No markup on AI usage
What is an AI DM setter?
A setter's job, done by an AI agent inside your DMs
In sales, a setter is the person who turns interest into appointments. They don't close the deal; they start the conversation, find out whether the lead is a fit, and get a qualified person onto the calendar of whoever does close. Coaches, agencies, course creators, clinics and high-ticket service businesses have hired setters for years, because the gap between "I'm interested" and "I'm booked" is where most leads are lost.
An AI DM setter does that job inside your direct messages. When someone replies to your story, asks a question under a reel and then DMs you, messages your Facebook page, or texts your WhatsApp Business number, the AI DM setter reads the message, understands what the person actually wants, and replies like a well-briefed member of your team. It asks the qualifying questions you care about, one at a time, saves the answers to your CRM, and when the lead fits, offers real open times from your calendar and books the one they pick.
The point isn't to replace the conversation with a script. The point is that every lead gets a fast, relevant, human-sounding reply, every time, including the ones that arrive at 11pm on a Saturday, the ones that come in all at once after a post takes off, and the ones your team would have answered three days late.
How an AI DM setter is different from a DM automation bot
Most "DM automation" tools are keyword triggers. Someone comments "GUIDE" and gets a link; someone types "price" and gets a canned message with buttons. They're useful for delivering a lead magnet, and they fall apart the moment a real person writes a real sentence: "I saw your reel about meal prep, I'm a nurse on night shifts, would your plan work around that?" A keyword bot has nothing for that. A human setter does. So does an AI DM setter.
An AI DM setter runs on a large language model, so it reads the whole message, including typos, slang, voice-to-text rambles and three questions packed into one. It answers from your own knowledge base: your offer, your prices, your FAQs, your policies. It follows a playbook you design, so it stays on the path to a booking instead of chatting forever. And it can act: write a field, add a tag, move a pipeline stage, check a calendar, book, reschedule, cancel and hand the thread to a person.
What ReplySetter adds: a setter you can see inside
The hard part of putting AI in your DMs isn't getting it to talk. It's trusting it. ReplySetter is built around that. You design the setter's job on a visual flow canvas: steps with objectives ("find out their goal", "find out their timeline"), scenarios for the moments that go off-script (price objections, "is this a scam?", "can I talk to a real person?"), and the tools each step may use. You give it a character: a name, a voice, a model and guardrails that keep it on topic.
Then you test it in a sandbox, playing the lead yourself, and watch every thought, every knowledge-base lookup and every tool call in a full reasoning trace. When something's off, you fix the instruction and rerun the same turn. When it goes live, you choose how much it does alone: suggest every reply for approval, draft sends while lookups run on their own, or work fully on autopilot. Most teams start in suggest mode, approve for a week, and switch to auto once every approval is a yes.
Where the DMs come from
ReplySetter connects to Instagram, Facebook Messenger, WhatsApp and SMS through GoHighLevel, the CRM many coaches and agencies already run their pipeline on. Your social accounts and phone numbers stay connected in GoHighLevel; ReplySetter answers the conversations that arrive there, on the same channel the lead wrote on, so every reply shows up in the contact's GoHighLevel conversation where your team can see it. Calendars, custom fields, tags, pipelines and workflows all come from the same sub-account.
The same setter can also answer email (Gmail, Outlook and any IMAP mailbox), your website chat widget, and replies to cold email campaigns in SmartLead or Instantly. One playbook, one record of every conversation, whichever door the lead walked in through.
Why your DMs need a setter, and why it should be AI
DMs are where interest turns into revenue, or quietly doesn’t. These are the six ways leads slip away, and what an AI DM setter does about each.
Speed decides who gets the call
A DM is an impulse. The person who just watched your reel is interested now, and in ten minutes they're watching someone else's. A reply in seconds catches them while they're still in the app; a reply tomorrow morning reaches someone who's forgotten why they wrote.
Content comes in waves
When a post takes off you get 200 DMs in an evening, then a quiet week. A human team is either overwhelmed or idle. An AI DM setter gives the 200th message the same careful reply as the first, and costs almost nothing in the quiet week.
Good setters are hard to keep
Hiring, training, commissions, sick days, turnover and a new person every few months who doesn't know your offer yet. The AI setter learns your offer once, never forgets a detail, and never has an off day on the call that mattered.
Your calendar fills with the wrong people
Booking everyone who says "sounds cool" wastes your closers' time. The AI DM setter asks the questions that decide a good call (goal, timeline, budget, decision maker) and only offers times to leads that fit. The rest get a helpful answer and a nurture tag.
Most deals are lost to silence
"Let me check and get back to you" is where pipelines go to die. The setter follows up on its own, with spaced, context-aware nudges that build on the conversation, and stops the moment the lead replies.
You can't coach what you can't see
With human setters you find out a lead was mishandled when the lead tells you. With ReplySetter every reply has a reasoning trace, every field update is logged, and analytics show where conversations drop off, so you fix the step, not the symptom.
How to set up an AI DM setter in 8 steps
From connecting your channels to a setter that books calls on its own. Most of it is writing down what your best setter already knows.
1. Connect your DM channels
Connect your GoHighLevel sub-account with a Private Integration token or the Marketplace app. Instagram, Facebook Messenger, WhatsApp and SMS conversations in that sub-account become one ReplySetter inbox, along with its calendars, custom fields, tags, pipelines and workflows. Recent history syncs for context, and the setter only ever answers new messages, so nobody from last year gets a surprise reply.
2. Teach it your offer
Add what the setter should know to a knowledge base: your program or service, prices and payment plans, who it's for and who it isn't for, results and case studies, FAQs, refund policy, onboarding steps. Upload documents or paste text. The setter searches it before answering questions about your business, so it quotes your real terms instead of inventing them.
3. Give it a character
Pick the name it signs as, the tone (warm and casual for Instagram, a little more formal for WhatsApp business enquiries), the AI model it runs on and up to five backup models in case a provider has an outage. Lock it to the languages you sell in, and keep its guardrails on so it politely refuses homework help, coding requests and anything else that isn't your business.
4. Design the setter's playbook
On the flow canvas, lay out the conversation as steps with objectives: answer the question, find out their goal, timeline and budget, offer a call, confirm the booking. Add scenarios for the moments that go off-script, such as price objections, "is this legit?", "I need to ask my partner" and requests for a human. Mention @Field to save answers, @Calendar to book, and the setter knows which tool to use where.
5. Decide who it answers and when
Use tag filters so the setter only answers new leads, or never answers existing clients. Set business hours and a timezone per inbox if you want it to reply only during the day, or leave it on around the clock. Choose which conversations get follow-up sequences and how far apart the nudges are.
6. Test it in a sandbox
Open a test session and play the lead: the curious one, the skeptic, the one who rambles, the one who goes quiet. Watch each reply's full trace: what the setter thought, what it looked up, which tools it called and what they returned. Rate replies, fix the instructions that produced bad ones and rerun the same turn until it's right. Save good conversations as tests to rerun after every change.
7. Go live in suggest mode
Switch it on with every reply waiting for your approval. You'll see each suggested message in Approvals, with the reasoning behind it, and approve, edit or reject it in a click. In draft mode, tags, field updates and lookups run on their own and only sends wait for you.
8. Hand it the keys and watch the numbers
When a week of approvals is all green, move to full auto. Analytics show response times, booked calls, where conversations drop off and which steps lose people. A/B test a new opening message or a different qualifying question against the current one, and keep whichever books more calls.
Everything a great setter does, in every DM
A closer look at how the AI DM setter answers, qualifies, books, recovers and knows when to step aside.
Instagram, Facebook, WhatsApp and SMS, answered by one setter
Your leads don't pick one channel, so your setter can't either. Someone finds you on Instagram, messages your Facebook page a week later, and texts the number on your booking page when they're ready. With ReplySetter connected to GoHighLevel, all of those conversations reach the same AI DM setter, which replies on the channel the lead used, in the thread your team already sees in GoHighLevel.
Real conversations are messy, and the setter is built for that. Three messages sent in a row get one thoughtful reply instead of three rushed ones. Emoji-only messages and reactions are recognised and don't trigger a pointless answer. A safety sync runs behind GoHighLevel's webhooks, so a lost notification never means a lost lead.
- Instagram DMs, Facebook Messenger, WhatsApp, SMS, email and live chat
- Replies go out through GoHighLevel, on the lead's own channel
- Message bursts merged into one reply; reactions ignored
- Business hours and timezone per inbox, or always on
Qualifies every lead before it touches your calendar
Give each step objectives, the things the setter must find out before moving on: what the lead wants, how soon, roughly what they can invest, whether they make the decision. The setter asks naturally, one question at a time, weaving each into the conversation instead of firing a form at the lead. When a lead volunteers an answer early ("I need this before my wedding in June"), it's noticed and not asked again.
Every answer is written to the right GoHighLevel custom field as it's given, so your closer opens the call already knowing the goal, the deadline and the budget. Dropdown and checkbox fields only receive valid options. Leads who aren't a fit get a helpful answer, a tag and a route to your nurture sequence, instead of a call that wastes everyone's time.
- Objectives per step: the setter won't offer times until they're met
- Answers saved to custom fields with an @Field mention
- Unqualified leads tagged and routed, not booked
- Pipeline stage moved by name as the lead progresses
Books the call on your real calendar, then confirms it
When a lead qualifies, the setter checks real availability and offers two or three specific open times, never a vague "when works for you?". It books the slot the lead picks, confirms it in the thread, tags the contact and can drop them into your reminder workflow. If they need to move it, it finds the existing appointment and reschedules; if they cancel, it frees the slot. No double bookings, no back-and-forth.
Use GoHighLevel calendars, or connect Cal.com and Calendly directly. With Cal.com or Calendly, the setter can also send a booking link, and bookings made on the page reach the conversation. After a no-show or a cancellation on the booking page, the setter can message the lead to find a new time, so a missed call becomes a rebooked call instead of a dead lead.
- Check availability, book, reschedule, cancel and look up appointments
- Different calendars per situation: discovery call here, consult there
- Cal.com and Calendly with booking links and booking-page sync
- No-show and cancellation recovery messages
Handles "too expensive" and "is this legit?" without losing the thread
Scenarios are the setter's playbook for the moments that derail a conversation. Define one for price objections, one for trust ("is this a scam?"), one for "I need to talk to my partner", one for "I'll think about it". Whatever step the conversation is on, when the moment comes the scenario takes over with your instructions, answers from your knowledge base (the payment plan, the guarantee, the case study that matches), then returns the conversation to where it was.
Because the answers come from your documents, the setter only offers terms you actually have. It won't invent a discount, promise a result you don't promise, or quote last year's price. If a question isn't covered, you decide whether it says it'll check with the team or hands the thread to a person.
- Scenarios for objections, trust questions and stalls
- Answers grounded in your knowledge base, with sources in the trace
- Returns to the step it left, so the booking stays on track
- Stays on topic: off-topic requests are politely declined
Follows up when they go quiet, and stops the moment they reply
Most leads who say "let me get back to you" never do, unless someone gently checks in. Follow-up sequences send spaced nudges when a lead goes quiet, each one written fresh from the conversation so far, not a copy-pasted "just bumping this". The first might offer the slot that's still open, the second a different week, the last a friendly door left open.
The moment the lead replies, the sequence stops and the setter picks the conversation up with full context. You choose the delays and how many nudges each conversation gets, and you can keep them inside each platform's messaging rules (more on that below).
- Spaced nudges with delays you choose
- Every nudge written from the conversation, never generic
- Stops automatically when the lead replies
- Paused contacts and opted-out leads never get nudged
Knows when to step aside for a person
Some conversations should go to a human: the lead who asks for one, the big-ticket enquiry, the complaint, the question nobody wrote down. Tell the setter when to hand off and to whom. It assigns the conversation to a GoHighLevel user, leaves a short summary of what it learned, and pauses itself for that conversation, so your teammate never gets interrupted mid-reply by the AI.
When a person replies from GoHighLevel themselves, the AI can pause automatically for that conversation too. Turn it back on when you're done and it picks up from the current state, not from where it left off. If a lead asks to stop, the setter opts them out, and they hear nothing more from it.
- Hand off to a named GoHighLevel user with a summary
- AI pauses per conversation; resume any time
- Pauses when your team replies manually
- Opt-out handled the moment a lead asks
Replies at 9:47pm, not tomorrow at 8:10am
DMs peak in the evenings and at weekends, when people are scrolling, which is exactly when most teams are offline. The AI DM setter replies within seconds whenever the message arrives, and handles a sudden wave of DMs after a post goes viral as calmly as a single one. Messages are paced and retried behind the scenes, so a busy evening never trips a provider's rate limit or drops a reply.
If you'd rather it only replied during business hours, set hours and a timezone per inbox and it holds messages until you open. Many teams do the opposite: the setter covers nights and weekends, and the humans take over during the day.
- Seconds from DM to reply, day and night
- Waves of DMs paced and queued, never dropped
- Up to five backup AI models if one provider fails
- Business hours per inbox when you want them
See every thought before a lead ever does
Every reply comes with a trace: the step it was on, the objectives it still needed, what it searched in the knowledge base and found, which tools it called with which inputs, and the token cost of the run. In a test session you play the lead and read along; in production the same trace sits beside every message in the inbox.
When a reply's wrong, you don't guess. You see the instruction that caused it, change it, and rerun the same turn to compare. Save the conversations that matter (the tricky objection, the lead who rambles) as tests and rerun them after every change, so improving one step never quietly breaks another.
- Full reasoning trace for every reply, in tests and live
- Rerun a turn after changing an instruction
- Saved test conversations to catch regressions
- Token cost of each run, so you know what a booking costs
AI DM setter vs human setter vs keyword DM bot
Three ways to work your DMs, and what each is good at.
| Compared on | AI DM setter | Human setter | Keyword DM bot |
|---|---|---|---|
| Reply time | Seconds, any hour | Minutes to hours; offline at night | Instant, for matching keywords |
| Understands free-text messages | Yes, including typos and rambles | Yes | Only the keywords it was given |
| Answers questions about your offer | From your knowledge base | From memory and training | Canned replies only |
| Qualifies leads | Objectives you define, one question at a time | Yes, when they remember to | Button menus |
| Handles objections | Scenarios grounded in your terms | Depends on the setter | No |
| Books, reschedules, cancels | On your real calendar | Yes, manually | Sends a link at best |
| Updates CRM fields and pipeline | As the conversation happens | If they find time | Rarely |
| Follows up when leads go quiet | Context-aware nudges, stops on reply | Inconsistently | Generic drip, if any |
| Handles a wave of 200 DMs | Same care for every one | Backlog for days | Yes, superficially |
| Shows why it said what it said | Full trace for every reply | No | No |
| Cost | From $97/month plus AI usage at cost | $2,000+/month or commission | $15–$100/month |
Which one do you need?
A keyword DM bot is the right tool for one job: delivering a lead magnet when someone comments a trigger word. It's cheap and instant, but it can't hold a conversation, so the leads who write a real question get nothing useful. Many businesses run one alongside an AI setter: the bot delivers the guide, and the setter answers whatever the person writes next.
A good human setter is still excellent at what they do, especially on high-stakes calls and in relationships that run over months. Their limits are hours, attention and consistency. The best results usually come from combining them: the AI DM setter replies to everything instantly, qualifies, books and follows up, and hands the conversations that need a person to a person, with a summary. Your human setters spend their day on the leads worth their time.
Who uses an AI DM setter
Any business where conversations in DMs turn into calls, consultations or appointments.
Coaches & consultants
Answer "how does your coaching work?" in seconds, qualify goal and budget, and book discovery calls straight from story replies and DMs.
Agencies
Run a setter for every client from one account. On Agency, one setter can serve many GoHighLevel sub-accounts, matching each client's calendars, pipelines and fields by name.
Course & community creators
Turn launch-week DM floods into enrolment calls, answer curriculum and payment-plan questions, and follow up with the people who said "maybe next round".
Med spas & aesthetics
Answer treatment and pricing questions from Instagram, check suitability, and book consultations without tying up the front desk.
Fitness & gyms
Book trial classes and personal-training consults from Instagram and WhatsApp, on separate calendars from the same conversation.
High-ticket offers
Filter for budget and readiness before a closer's time is spent, and hand the biggest opportunities to a person with a summary.
Real estate & mortgages
Reply to Facebook and Instagram enquiries instantly, qualify timeline and financing, and book viewings or calls.
Local & home services
Turn "how much for…?" DMs into booked estimates, saving the job type and address to the contact as the lead types.
AI DM setter scripts: three playbooks to copy
Each maps straight onto a flow: one step per line, with scenarios for the moments that go off-script.
Story reply to discovery call (coach)
For warm leads replying to content. Four steps, two scenarios.
- Answer: thank them for the reply, answer their actual question in two sentences from the knowledge base.
- Goal: ask what they'd most like to change in the next 90 days; save it to @Field.goal.
- Fit: ask what's stopped them so far and how soon they want to start; save timeline.
- Offer: if goal and timeline are set, offer two open times on @Calendar.Discovery Call and book the pick.
- Scenario, price: explain the payment plan from the knowledge base, then return to the offer.
- Scenario, not now: tag nurture and add them to the follow-up sequence.
Inbound enquiry to consultation (clinic or med spa)
For first-time enquiries asking about a treatment or price.
- Answer: give the starting price and what's included, from the knowledge base.
- Suitability: ask the screening questions you need, one at a time; save each to its field.
- Route: if a screening answer rules them out, explain kindly and hand off to the clinic team.
- Offer: otherwise offer two consultation times and book the chosen one.
- Confirm: send the address and preparation notes, tag booked, start the reminder workflow.
- Scenario, reschedule: find the appointment and move it, without a second booking.
Launch-week DM flood (course creator)
For a burst of DMs after a launch post. Speed and consistency matter most.
- Answer: reply to the specific question (curriculum, schedule, price) from the launch FAQ.
- Readiness: ask whether they're looking to join this round; save the answer.
- Offer: for ready buyers, offer an enrolment call or send the checkout page from the knowledge base.
- Scenario, payment plan: explain the instalment option and its terms.
- Scenario, ask a person: hand off to the launch team with a summary.
- Follow-up: two nudges before cart close, none after it.
10 rules for an AI DM setter that books more calls
The model matters less than the brief. These habits make the biggest difference to how many DMs become booked calls.
1 Answer first, qualify second
A lead who asked a question and got a qualifying question back feels processed. Have the first step answer what they actually asked, in two or three sentences, then ask your first question. Trust first, then information.
2 One question per message
Stacked questions ("What's your goal, budget and timeline?") read like a form and get partial answers. Make each qualifying item its own objective, and the setter will ask them one at a time, in a natural order.
3 Write like you DM, not like you email
Instagram DMs are short, casual and quick. Tell the character to keep replies to two or three short sentences, skip formal greetings and sign-offs, and mirror the lead's tone. Save the longer, more formal voice for email inboxes.
4 Offer specific times, never "when works for you?"
An open question about availability starts a back-and-forth that loses leads. Two or three specific open times make it a one-tap decision. The setter checks your real calendar, so every time it offers is genuinely free.
5 Put your real terms in the knowledge base
The setter is only as accurate as what you give it. Write down prices, payment plans, guarantees, who the offer isn't for, and the answers to the ten questions you get most. Update the document when the offer changes, and the setter changes with it.
6 Write scenarios for the moments you dread
Think of the messages that make a new setter freeze: "is this a scam?", "that's too expensive", "I need to ask my husband", "just send me the price". Give each a scenario with how you'd like it handled. These few scenarios make the biggest difference to booked calls.
7 Be honest that it's an assistant
Give the character a name and personality, but don't have it claim to be a specific human. If a lead asks whether they're talking to a bot, it should say it's an assistant for your business and offer a person. Honesty here protects your brand and keeps you on the right side of platform rules.
8 Decide your hand-off rules up front
Write down when a person should take over: the lead asks, the deal is above a size, there's a complaint, a medical or legal question. Make each a scenario that hands off to a named user. Leads trust an AI setter more when they know a human is one message away.
9 Start in suggest mode for a week
Approve every reply for the first week. You'll learn more about your own sales conversations than you expect, and every edit you make is a hint for an instruction to improve. When a day passes without an edit, move to draft, then to full auto.
10 Test one change at a time
When you want to try a new opener or a different qualifying question, A/B test it against the current version instead of swapping it in. Each lead sticks to one variant, and analytics show which one books more calls.
Staying inside platform rules, and on brand
An AI DM setter should protect your accounts and your reputation, not put them at risk. Here’s how.
Messaging windows on Instagram, Messenger and WhatsApp
Meta's platforms give businesses a window, generally 24 hours after the lead's last message, to reply freely. Outside it, Instagram and Messenger restrict promotional messages and WhatsApp requires approved templates. Set your DM follow-up delays to land inside the window, and move longer-term follow-up to SMS or email where the lead has opted in. Check Meta's current policies for the details that apply to your account.
No cold DMs, no scraping, no spam
ReplySetter answers people who wrote to you. It doesn't send cold DMs to strangers, scrape followers or mass-message lists, which is what gets accounts restricted. Every conversation starts with the lead, which is also why the replies land so well.
Opt-outs honoured immediately
When a lead says stop, the setter opts them out in GoHighLevel and stops replying and following up. Tag filters let you exclude existing clients, partners or anyone else who should never hear from the AI.
On topic, on brand, no made-up promises
Characters stay on your business: requests for homework, code, essays or anything unrelated are politely refused, and repeated misuse stops the AI for that conversation. Answers come from your knowledge base, so the setter can't invent a discount or a guarantee you don't offer.
Your keys, your data, your models
You connect your own AI provider key (Anthropic, OpenAI or OpenRouter), so usage is billed by your provider at cost with no markup, and you choose the model. Add backup models so a provider outage never leaves DMs unanswered.
Every action on the record
Every reply, field write, tag, booking and hand-off is logged with the reasoning behind it. If a lead ever asks why they got a message, or a client asks what the setter has been doing, the answer is one click away.
What answering every DM is worth
Every business's numbers are different, so here's the shape of the maths rather than a promise. Plug in your own figures.
| Example business | Figure |
|---|---|
| Inbound DMs a month | 300 |
| Replied to within a few minutes today | 40% (evenings and weekends missed) |
| Replied to within seconds with an AI DM setter | 100% |
| Qualified and offered a call | 30% of replied leads |
| Booked calls a month, today | 36 |
| Booked calls a month, with every lead answered | 90 |
| Close rate on calls | 20% |
| Extra clients a month | about 11 |
Even if the AI setter only closes half that gap, the extra calls come from leads you're already paying to attract with content and ads. Against that, the cost is a plan from $97 a month plus AI usage, which for a DM conversation is usually a few cents. The debugger and analytics show the token cost of every run, so you can work out your own cost per booked call.
How to choose an AI DM setter: 10 questions to ask
Whichever tool you pick, ask these first. Here’s how ReplySetter answers them.
| Question | Why it matters | ReplySetter |
|---|---|---|
| Does it understand free-text DMs, or only keywords? | Most leads write real sentences, not trigger words. | Full language model understanding, any phrasing |
| Does it answer from my own documents? | Made-up prices and promises cost you sales and trust. | Knowledge bases, with sources in every trace |
| Can it book directly on my calendar? | Sending a link and hoping loses people. | GoHighLevel, Cal.com and Calendly: book, move, cancel |
| Does it update my CRM as it goes? | Your closer needs the answers before the call. | Custom fields, tags, pipeline stages, workflows |
| Can I see why it said what it said? | You can't improve or trust a black box. | Full reasoning trace for every reply |
| Can I approve replies before they go out? | You need a safe way to go live. | Suggest, draft and full auto, per agent |
| Does it know when to hand off to a person? | Some leads should always reach a human. | Hand-off to a named user, AI pauses itself |
| Will it follow up, and stop when they reply? | Most bookings come after the first conversation. | Context-aware sequences that stop on reply |
| Can I test changes before leads see them? | A good-looking edit can break another step. | Sandbox sessions, reruns and saved tests |
| Is AI usage marked up? | Per-message markups add up fast at volume. | Your own keys, billed at cost by your provider |
AI DM setter pricing
Build it free. Put it in your DMs from $97 a month. AI usage is billed by your own provider at cost, with no markup.
Free · $0
Build and test your setter's playbook, character and knowledge base, and try it on Gmail, Outlook or a website chat.
- 1 agent, 1 inbox
- Test & Debug with full traces
- Suggest and draft autonomy
Basic · $97/mo
Connect GoHighLevel and put the setter on Instagram, Facebook, WhatsApp and SMS DMs.
- 5 agents, 5 inboxes
- Booking, CRM actions, follow-ups
- Full auto, A/B testing, analytics
Agency · $297/mo
Run DM setters for many clients, with one agent serving many GoHighLevel sub-accounts.
- 50 agents, 25 inboxes
- Many sub-accounts per agent
- MCP server, priority support
Every plan includes the flow builder, characters and Test & Debug. Compare plans in full
AI DM setter: frequently asked questions
What is an AI DM setter?
An AI DM setter is an AI agent that does a sales setter's job inside your direct messages: it replies to people who DM you on Instagram, Facebook Messenger, WhatsApp or SMS, answers their questions, qualifies them with the questions you choose, and books a call or appointment on your calendar when they're a fit.
Which channels does ReplySetter's AI DM setter work on?
Instagram DMs, Facebook Messenger, WhatsApp and SMS, through a connected GoHighLevel sub-account, plus GoHighLevel live chat and email. Facebook Messenger also works through HubSpot's conversations inbox. The same setter can also answer Gmail, Outlook, IMAP mailboxes, your website chat widget, and SmartLead or Instantly replies.
Do I need GoHighLevel?
For Instagram, WhatsApp and SMS, yes: ReplySetter connects to those channels through GoHighLevel, where your social accounts and numbers are connected. Facebook Messenger works through GoHighLevel or HubSpot's conversations inbox. If you don't use GoHighLevel yet, you can still build and test your setter on the Free plan, and use it on email and website chat.
Will it sound like a bot?
Not if you brief it well. You choose its name, tone and model, keep replies short for DMs, and have it answer the lead's actual question before asking anything. Every reply is written fresh from the conversation, never from a template. Test it in the sandbox until it sounds like you.
Does it pretend to be human?
It shouldn't, and we recommend against it. Give the character a friendly name, but if a lead asks whether they're talking to a bot, it should say it's an assistant for your business and offer a person. A human is always one hand-off away.
Can it send cold DMs or message my followers?
No. ReplySetter answers people who wrote to you and follows up on those conversations. It doesn't send cold outreach DMs, scrape followers or mass-message lists, which is what gets accounts restricted.
Does it work with comment-to-DM automations?
ReplySetter doesn't trigger DMs from comments itself. If you use a comment automation to send a lead magnet, the setter can answer whatever the person writes back in the DM, with the conversation history in context.
How does it book calls?
It checks real availability on your GoHighLevel, Cal.com or Calendly calendar, offers two or three specific open times, books the one the lead picks and confirms it. It can also reschedule, cancel and look up existing appointments, and with Cal.com or Calendly send a booking link.
Can I approve messages before they're sent?
Yes. In suggest mode every reply and action waits in Approvals; in draft mode lookups, tags and field updates run but sends wait for you; in auto mode the setter works on its own. You can change the level at any time.
What happens when a lead asks for a real person?
The setter hands the conversation to the GoHighLevel user you chose, leaves a summary of what it learned, and pauses itself for that conversation. When your teammate replies, the lead sees it in the same thread. You can turn the AI back on whenever you like.
Will it reply to my old conversations when I connect?
No. Recent history syncs so the setter has context, but it only replies to new inbound messages.
How much does an AI DM setter cost?
Building and testing is free. DM channels through GoHighLevel are on Basic at $97 a month, and Agency is $297 a month for many clients. AI usage is billed by your own provider at cost, with no markup, and a typical DM reply costs a few cents.
Which AI model does it use?
The one you pick. Connect your own Anthropic, OpenAI or OpenRouter key and choose from hundreds of models, with up to five backup models per character if your first choice is unavailable.
Can it reply in other languages?
Yes. By default it answers in the lead's language. On paid plans you can lock a character to the languages you sell in, so it always replies in one of those.
Can an agency run DM setters for clients?
Yes. On Agency, one setter agent can serve many GoHighLevel sub-accounts, matching each client's calendars, pipelines, stages and workflows by name, and custom values resolve per sub-account.
How do I know it's working?
Analytics show reply times, booked calls, outcomes and where conversations drop off, and an AI conversation map shows the stages leads move through. Every reply has a full trace, and A/B tests tell you which version of a step books more calls.
More about the pieces behind it
Put an AI DM setter in your inbox today
Connect your DM channels, teach it your offer, test it in a sandbox and go live with exactly the autonomy you're comfortable with.