HookSendOne moment
Features
All 112of them, described properly. Comments, story replies, mentions, lives, ads and DMs answered on their own, an AI that replies from your own documents, moderation under your posts, a shoppable page for your bio, and a connector that lets your own AI run the lot. All of it on Instagram's official API.
Everything that can set an automation off. Instagram only lets a business message someone who contacted it first, so every one of these starts with something the other person did.
Someone comments your keyword under a post or reel and gets a DM from you within seconds, carrying a link, a discount code, a lead question or a row of product cards. Use it on the post where you would otherwise have typed "check my bio" a hundred times. One campaign can hold several keywords, or answer any comment at all.
A campaign can point at one specific post, at every post and reel on the account, or at the next reel you publish. The last one is for a launch you have not posted yet: you set the campaign up in advance, and it attaches itself to the reel the moment it goes live.
When someone replies to your story with a keyword, HookSend answers in the DM automatically. Point the campaign at one story or at every story you post. A campaign aimed at a single story pauses itself when that story expires twenty-four hours after it was posted, so a dead campaign never sits in your list looking live.
When somebody tags your account in their own story, HookSend can thank them, send a discount code or point them at a link. It is the cheapest way to reward the people already sharing your product, and it needs nothing from them beyond the mention.
A quick emoji tap on a story arrives as a reply with no words in it, which means a keyword can never match. Switch on "any reply" and the campaign answers those too, so a 🔥 gets the same link a typed keyword would.
When someone sends you a DM containing one of your keywords, such as PRICE or LINK, HookSend replies on its own. A comment campaign can listen to DMs as well, so "comment PRICE or DM me PRICE" gets both people the same answer from one campaign. The switch is called "Trigger on incoming DMs too" and it starts on for a new keyword campaign.
Keyword triggers explainedWhen a person shares one of your own reels or posts into your DMs, the campaign on that post answers as though they had commented on it. There is no switch: it is always on, because sharing a reel to ask about it is one of the most common ways a customer opens a conversation.
A Live campaign answers keyword comments during a broadcast, which is exactly when you cannot read them yourself. Choose "every live" to have it run on all of them, or "current or next live" to bind it to one broadcast: it attaches to the live already running when you save, or to the first comment on your next one, and then answers that broadcast only.
Comments left on an ad or a boosted post can be answered by their own campaign, separately from the organic post. Pick one ad from the list of ads your comments have actually arrived on, or answer comments on every ad. The ad list builds itself as comments come in, so a brand-new ad appears after its first comment.
Keywords are matched case-insensitively with emoji stripped, and any one of a campaign's keywords matching is enough. A campaign matches whole words, so LINK does not fire on "unlinked" or "linking". Knowledge bases and flow triggers use the same matcher and let you turn whole words off when you want a partial match. You can also drop keywords entirely and have a campaign answer every comment on the post.
The conversation a campaign sends, step by step. Every step below runs the same way whether the person arrived from a comment, a story reply, a keyword DM or a button tap.
Send a greeting first and the link second, with a button between them. The tap is what lets HookSend write again: Instagram allows a business only one private reply to a comment, and after that it may only message somebody who has answered. An opening with no button shares one bubble with the next step instead, and the greeting is what gets shortened if the two together run long.
A campaign's message can carry two link buttons, each with its own wording and its own destination. HookSend never sends your address directly; it sends a tracked link of its own, so every tap is counted and shows up in the campaign's statistics. Write {link} in the message body instead if you would rather the link sat inline in the text.
Send a swipeable row of up to eight product cards in one message, each with a photo, a title, a line of detail, a price and a button. Use it when the answer to "link?" is five products rather than one. Photos you upload are compressed in the browser before they go, and a card with no image falls back to a generated preview rather than arriving blank.
Paste a product URL — Amazon, Myntra, Flipkart, Zara, a Shopify store — and HookSend reads the page and fills in the title, the price, the photo and the merchant, with the button pre-labelled "Shop On" that shop. When a shop refuses to be read it says so and you add the card by hand, rather than saving something half-filled.
Add buttons whose taps behave exactly as if the person had typed a keyword. That is how a campaign passes somebody on: the button says whatever reads best, the keyword behind it belongs to a flow or a knowledge base, and the conversation carries on there. Instagram's template holds three buttons in all, link buttons included, and cuts a button's words at twenty characters.
Send the link only to people who follow you. Anyone who does not is asked to follow and given an "I'm following" button to tap once they have, which re-checks and then releases the link. When Instagram will not say whether somebody follows you, the gate asks rather than gives the link away — except after a tap, so nobody can get stuck asking forever.
Ask for an email address before the link, in your own words, and send a confirmation message once it arrives. The address lands on your Leads page against the campaign and the post it came from. A DM flow can collect a phone number instead, and either one counts as a lead.
Send a second message after the link has gone out: a nudge, a code, a question. You set the wait in minutes, and zero sends it straight away. Use it for the thing that should not share a bubble with the link, such as "did that work?" twenty minutes on.
HookSend can write alternative wordings of each message so a hundred people do not all receive one identical paragraph. Your own text stays exactly as you wrote it and is always one of the options; links and buttons are never touched. Which wording somebody gets is fixed to them rather than picked at random, so a retry or a second trigger never sends the same person a differently-worded version. The send log records which one they got, so you can see which wording earns the clicks.
As well as the DM, HookSend can reply to the comment publicly, so everyone scrolling sees that you answered. Give it several replies and it rotates between them rather than posting the same line fifty times under one reel. The public reply is delivered independently of the DM: if the DM fails the reply still goes out, and a retry never posts it twice.
Write {username} anywhere in a message and it becomes the person's name. Write {link} and it becomes your tracked link. When the link is going out as a button instead, the token is removed cleanly rather than left in the text, and any rewrite that would have lost a token is thrown away rather than sent.
Start from a pre-configured blueprint rather than an empty form, or paste a CSV with one row per campaign to set up many at once. Each imported row opens in the builder for you to check before it goes live, which matters when you are standing up thirty campaigns for a client.
Browse the campaign templatesAnswers written from your own documents, not from whatever a model happens to believe. Every AI action costs credits from your monthly allowance, and the cost of each one is published. Each of these has a page of its own with the whole thing demonstrated.
All the AI features, in depthGive HookSend a price list, a brochure or a service menu and it answers DM questions from it, in the language the customer wrote in. When your documents do not cover a question it sends your fallback message, or nothing, rather than inventing an answer.
Three ways to get your words in: upload a PDF or a text file up to 10 MB, paste the text straight in, or give HookSend the address of a public web page and let it fetch the content. There is no limit on how many documents one knowledge base holds.
A box on the Knowledge page runs a question through exactly what an inbound DM runs, and opens up the passages it read with a match score against each, so a wrong answer traces back to the paragraph that caused it. It costs no credits.
A panel over the last thirty days: people helped, answers sent, questions it could not answer, and chats open right now. The "could not answer" number is a list of what your documents are missing, which is the only maintenance a knowledge base really needs.
One switch, and any DM that no campaign, flow or keyword already handles is answered from your live knowledge bases. Use it when your inbox is mostly the same five questions; when your documents do not cover the message, nothing is sent.
Once a keyword opens a conversation, every message that person sends is answered without them typing the keyword again, because nobody writes "price" before each sentence. It closes itself after thirty quiet minutes, or at once if they say stop, end, cancel or unsubscribe.
Under each answer, HookSend can offer four or five follow-up questions your documents can actually answer, as taps. They come from a list written across all your documents, so a suggestion can point at any of them, and tapping one sends the full question.
Give a knowledge base up to three link buttons and every answer carries them, which is the difference between a help desk and something that sells: the AI explains the offer, then hands over a button to the checkout. Taps are counted per button.
Describe how you want replies to read, such as "casual, short, a bit of Hinglish", or leave it empty to have the AI match whatever language and tone the customer used. The first reply can open with a greeting using the person's name, written to fit the question rather than a stock line.
The moment you answer somebody yourself, from the Instagram app or the HookSend inbox, the AI goes quiet with that person for an hour by default, so it never talks over you. It applies to that one person, not the whole account.
Switch this on and an answer arrives as two or three short DMs, the way a person texts, with the suggestions on the last one. Each counts against your monthly DM allowance, which is why HookSend asks first.
Have the AI write or rewrite a campaign's DM copy and public reply while you build it. It writes in the same language and script as your original and never translates it, and a rewrite that loses a link or a placeholder is discarded rather than shown to you.
Paste the link to one of your own posts or reels and HookSend proposes a whole campaign: name, keywords, the DM and the public reply, filled into the builder for you to read and save yourself. Metered by campaign rather than by click, so regenerating inside one campaign costs nothing more.
Every AI action spends credits from a monthly allowance, and a credit is only ever spent on a real model call. The free rules throw out about three quarters of comments before anything is asked, which is why a small allowance covers far more comments than it looks like it should.
A flow is a conversation in steps. Build it as a simple list or on a canvas, publish it when it reads right, and it answers its keywords from then on.
Every flow can be edited two ways over the same steps. The simple view is a list, quickest for a short conversation. The canvas is full screen and drag-and-drop, where each choice is a line you can follow, which is what you want once a flow branches. A flow stays paused until you publish it, so nothing you are half way through building reaches a real person.
A step can send a picture or a video rather than text: the size chart, the venue, the thirty seconds that answers the question better than a paragraph. Upload it and the flow sends it as part of the conversation, in order, before whatever comes next.
The builder saves as you work and says so, so a closed tab costs nothing. A phone preview beside it shows what the conversation actually looks like in Instagram. And when a flow cannot go live, the reason is a sentence about the step rather than an error code, so you know which card to open. A live flow you edit into a broken state pauses itself rather than half-answering people.
A collect step asks a question and checks the answer before moving on: an email address, a phone number, a number, a date, a web address, free text, or a follow check. Anything that fails the check is asked again rather than saved wrong, and an email or a phone number lands on your Leads page as a lead.
Give a step up to three buttons, which stay in the thread and can be tapped later, or up to thirteen quick replies, which read as chips and vanish once the person answers. A cards step sends a swipeable row of up to ten cards with pictures. A card's buttons can open a page or move the conversation on, and the flow only waits for a tap when a button actually leads somewhere inside it.
Send the conversation one way or another on what you already know: an answer that equals or does not equal something, contains a word, exists at all, or a tag the person carries. Either branch can simply end the run. Combined with tags, this is how one flow serves an existing customer and a new one differently.
An AI route step reads what the person wrote and chooses between up to eight paths you have described, which is what you want when the answer will not fit on buttons. You name a fallback path, and the conversation goes there when nothing fits, when the model cannot decide, or when your credits have run out, so the chat never simply stops.
Branch on the clock: one path during the hours you keep, another outside them. You pick the days, the start and end times on a twenty-four hour clock, and the time zone by name, because "9 to 6" means nothing without one. The end time is exclusive, so 10:00 to 18:00 covers 17:59 and not 18:00, and hours that cross midnight work as you would expect.
A tag step puts a label on the person or takes one off, and sends nothing: the conversation carries straight on. Tags are what the Leads page filters by and what a condition in any other flow can read, so somebody tagged "hot lead" on Monday is still tagged on Friday. Spelling is normalised to lower case, so "Hot Lead" and "hot lead" are one label.
Send everything the flow has collected, as JSON, to an address you own: a Google Sheet through Apps Script, Zapier, Make, or your own site. Every delivery is signed so your receiver can prove it came from HookSend. The conversation carries on whether or not your address answers, because a Zap being down must never cost a customer their reply.
End the automated part and let a person take the thread, with a private note for whoever picks it up, such as "refund request". It is deliberately the end of the run: once a human is involved, a bot resuming mid-conversation is worse than one that stays quiet.
Every step carries a count of the people who got that far, so you can see where a conversation is losing them. It is usually the question that asks for one thing too many. Read it on the canvas next to the step itself rather than in a separate report.
7 ready-made flows you can open, read and change rather than starting from an empty canvas: AI front desk, Giveaway entry, Price list, Free guide, Open or closed, Order support, Book a call. Each is a working conversation with its steps already wired, which is the fastest way to understand what a flow can do.
Abuse, spam and scam links hidden within seconds, and genuine questions left where they are. Guard reads the comments on your posts and reels; your DMs are handled by campaigns, flows and the AI. The pattern rules cost nothing on any plan, and the AI checks come with the plans that include credits.
Every comment on your posts and reels is read the second it lands, and the ones that break your rules are hidden, deleted, held for review, or explicitly kept. Use it on the reel that brought in the wrong crowd, or leave it running so a buyer never sees the scam link under your best post. Every decision is recorded with the rule that made it, so nothing disappears on you quietly.
Try Comment GuardGuard runs in two passes with your campaign between them. The free pattern rules go first and stop obvious spam from ever earning a DM. Your campaign then replies at full speed. The AI rules run behind it and hide or delete a few seconds later, so moderation never slows down the message the real customer is waiting for.
Alongside the built-in rules you can write three kinds of your own: a word list, a regular expression for something with a shape to it, or a plain-English instruction the AI judges each comment against. A words rule is how competitor names and the phrases you are tired of get handled. Rules can be reordered and any one can be switched off without deleting it, up to sixty per account.
No model needed: schemed and bare links, obfuscated ones written "shop (dot) com", a list of thirty-three link shorteners and chat hosts that are always suspicious, email addresses including the "(at)" form, phone numbers left to be contacted on, too many emoji, adult emoji, tagged handles and stuffed hashtags. Instagram's and Facebook's own domains are never treated as spam, and your campaign hashtags can be excluded from the count.
When several different accounts post the same comment under your post inside five minutes, that is a bot run rather than a coincidence. Switch the rule on, set how many duplicates it takes, and they go together. It is off by default because a genuinely catchy comment gets copied too.
A rule is a router, not only a bin. A keep rule explicitly lets a comment through and flags the person as a lead, which is how "where do I buy this" survives a broader rule sitting underneath it. Guard's feed shows the comments it kept as well as the ones it took down, because a moderation tool that only shows its own removals gives you no way to tell quiet from broken.
The AI rules score each comment out of a hundred. At or above your threshold Guard acts on its own; below it the comment waits for you rather than being guessed at. The threshold starts at 80. Every AI rule is scored before anything is decided and the heaviest call wins whatever the rule order, so a 99% impersonation match is never beaten by a 55% rudeness match.
Tell Guard what your account is about and the AI rules read comments in that light, so "fake" under a dupe-spotting page means something different from "fake" under a skincare brand. It is prefilled from your storefront when you have one, and it is the single cheapest fix for a rule that keeps getting your own audience wrong.
Before any model is asked, a comment made entirely of ordinary words — praise, prices, shipping, the Hinglish equivalents — is let through with no call at all. One unfamiliar word anywhere and the whole comment is judged. It is deliberately not a length rule, because abuse is usually shorter than praise, and it is why a small credit allowance covers many times as many comments as it looks like it should.
Guard reads every new comment and records exactly what it would have done while hiding and deleting nothing, under its own tab. Run it for a day, two days or a week before you trust it with your audience. Your DM campaigns keep running the whole time: watch only holds back the moderation, never the selling.
Type a comment and watch it go through the same function the live webhook calls. You get the verdict, the language it detected, and a numbered trace of every rule in the order it ran with a plain-words reason each — "no link", "12 emoji, under your limit", "91%, at or above your 80% line". There are ready-made samples to press, deliberately multilingual, and a rule can be tested while you are still typing it.
The honest test: a read-only pass over your real history, up to fifteen posts and five thousand comments across the last ninety days, which hides and deletes nothing. You get four counters for what it would have hidden, deleted, asked you about and left alone, a chart of which rule caught what, and the calls Guard was least sure about first, because those are the ones it is getting wrong on your actual audience.
Tabs for all, hidden, deleted, waiting on you, let through and watch only, with counts, a search across comment text and handles, and a chart of what Guard took down in the last twenty-four hours grouped by rule. From any row you can put a comment back, hide it, delete it, block the handle, or mark that Guard got it wrong without moving the comment.
Hidden is not gone: the person who wrote it still sees their own comment, so they get no signal, and one tap puts it back. Deleted really is gone, because Instagram never returns a deleted comment, which is why Guard keeps the wording of a deleted comment for forty-eight hours so you still have the handle, the words and the time if you want to report somebody.
A panel listing the handles with more than one comment hidden or deleted in the last thirty days, with the count, which is what the block button is for. Blocked handles are kept on their own list and read before any other rule runs, so a repeat offender is dealt with once rather than re-caught every week.
Guard skips people it already knows: anyone on your Leads page and anyone you have messaged before, both on by default. Your own comments under your own posts are skipped too, since running the rules over your own words costs a check and decides nothing. Verified accounts can be protected as well, which is off by default.
Three lists that always win: words that must never trigger a rule, handles whose comments are never touched, and link hosts never treated as spam. Use the host list for your own shop, so a fan quoting your product page is not filed as a spammer. The lists are checked before the rules, so nothing on them ever reaches a model.
Guard watches from the day you switch it on, which leaves the mess already sitting under your older posts. Load one of your last thirty posts, have up to five hundred of its comments scored against your live rules, and see what would go. Everything matched arrives ticked and you untick what you want kept, then choose once whether the sweep hides or deletes.
The sweep runs in the background with live progress, so you can close the page and it keeps going, and it tells you how many Instagram refused to touch. Each comment gets a normal archive row labelled as a clean-up, so a sweep shows up in your history exactly like anything Guard did on its own. One sweep at a time per account.
If a rule errors, a model is unavailable or Instagram refuses the call, the comment is left alone and your campaign runs regardless, and the failure is written into the trace rather than acted on. A moderation failure should cost at most an unmoderated comment, never a customer who commented and got nothing.
A public page your Instagram bio can point at, built from the posts and links you already have.
A public page at your own username with your photo, your display name, your bio, your links and your posts on it, plus a row of social links for Instagram, YouTube, WhatsApp, a website, email, X and TikTok. Active campaigns sync their posts onto it automatically, and a post you take off stays off rather than reappearing the next time the page loads.
A theme changes the structure as well as the colours: how cards are treated, how sharp or round the corners are, whether your avatar is a circle, a squircle or a square, and whether the shop reads as a grid, an editorial column or a stack. Each one previews as a live miniature so you pick by looking rather than by name.
Add as many links as you like, drag them into the order you want, highlight the one that matters this week, and switch one off without deleting it. Each link keeps its own tap count, so you can see which of them your followers actually use.
Tag products onto a post on your storefront: a picture, a title, a subtitle, a price with the old price struck through, and the shop it comes from. Somebody browsing your grid taps a post and gets the exact products in it. Each product counts its own taps.
Reading and answering the conversations your automations started, and finding the ones worth money.
Read your Instagram conversations inside HookSend and reply by hand when a real person needs to. Replying here is what pauses the AI with that person, so you never end up talking over your own bot. Threads are filtered by connected account.
Instagram only lets a business reply within twenty-four hours of somebody's last message. Each thread says whether the window is open and roughly how many hours are left, counting down as you read it, so you find out before you type rather than after Instagram refuses the send.
Set the questions Instagram shows as tappable buttons when somebody opens a chat with you for the first time, before they type anything. The number of boxes and the length allowed come from Instagram itself rather than being guessed, and after you save, HookSend reads back what Instagram actually has rather than trusting what it sent. Empty a box to take that question down.
Paid and barter approaches are pulled out of your inbox and listed on their own screen with the terms already read out of the message: the brand, whether it is the brand or an agency writing, the fee, what they want ("1 reel", "2 stories"), and any deadline. Detection deliberately does not look for the word "collab", which catches fans saying "loved your collab" and misses the message worth money.
Each approach carries a score for how likely it is a genuine collaboration and a separate score for the pay-a-fee-to-be-featured kind, so the obvious traps sort themselves to the bottom. Offers are typed as paid, barter, affiliate, an event or unclear, and you move each one through new, talking, agreed, passed or spam with your own private notes attached.
One summary a day of the brands that wrote to you, at an hour and a time zone you choose, defaulting to nine in the morning. It is only ever sent on days a brand actually wrote, so there is no "nothing happened" email. Turn the email off and keep the detection, or turn the detection off entirely and keep what has already been filed.
People already on your Leads page are skipped, because somebody who gave you their email through a campaign is a customer talking to you rather than a brand approaching you. That is on by default. Skipping everybody you have ever messaged is available too and off by default, since a brand that once got a campaign DM can still write with a real offer later.
A collaboration is a negotiation, so HookSend reads the offer and files it but never replies to it. What you get is the message, the terms and a place to keep your notes; the answer stays yours.
Sends, clicks, leads and reach, per campaign and per account, with a version you can hand to a client.
Sends, clicks, captured leads and your top keywords across the workspace, filtered by connected Instagram account. It is the screen that tells you whether last night's reel actually did anything, without opening a campaign.
Impressions and the unique accounts reached, likes, comments, saves and shares, across your last 25, 50 or 100 posts or all of them. Under it, an engagement chart over 24 hours, 7, 30 or 90 days or a range you pick, and a table of every post with views, reach, likes, comments, saves, shares and the date, each row linking out to the post itself.
Every link HookSend sends is a tracked link of its own, so clicks are counted per campaign and per link rather than guessed at from your site's analytics. Each click records when it happened, so you can see the shape of the hour after a post goes out.
Every campaign gets its own report at an unguessable address, and the Share button copies the link. The page carries DMs sent, skipped, failed, clicks, the click-through rate and a seven-day sent-against-clicks chart, with none of the workspace controls, so nobody can change a campaign from it. Useful when the person paying for the campaign should not have a login.
Every email address and phone number your campaigns and flows collected, with the person's Instagram handle, the campaign and the post it came from. Filter by campaign, connected account, post or tag, and download the result as a CSV to take into your mailer or CRM.
Every comment that matched, the keyword it matched, whether the DM was sent, queued, failed or skipped, and which wording the person received. Nine filters, and the skips say why: already answered, past Instagram's hourly rate, past your plan's allowance, or no keyword match. It is where you look when somebody says they never got their link.
A row that failed, was rate limited or hit your plan's allowance gets a retry link, and the result comes back inline. Use it after topping up DMs, or after an Instagram wobble, to give the people who missed out their message rather than writing the post off.
Views, reach, saves and shares for your posts and reels, next to the sends and clicks your campaign got from them. Seeing both in one place is what tells you whether a reel underperformed or whether the keyword was simply wrong.
Instagram keeps only about thirty days of account insights, so HookSend records your follower total once a day and keeps the history itself. Over a few months that becomes a growth chart nobody else can show you, with the daily change alongside the total.
Once a week, an email with the DMs sent, the clicks, the leads and your best campaign of the period, with a link to that campaign's shareable report. It only goes to workspaces that actually did something that week: a summary reading "0 DMs, 0 clicks" is a weekly reminder that nothing is working, which is not what it is for.
One workspace, several Instagram accounts, and colleagues who can see the work without being able to break it.
One workspace can run several Instagram accounts, with the dashboard, the logs, the leads and the campaigns all filtered by account. Campaigns share one workspace-wide pool rather than a quota per account, so a quiet client's unused campaigns are not wasted.
Invite colleagues into your workspace as admins or as members. The owner keeps billing. Admins connect Instagram accounts, build and edit campaigns, flows and knowledge bases, and invite other people. Members get read-only access, so a client or a junior can follow the results without changing anything.
If you have been invited into somebody else's workspace as well as running your own, a switcher next to the logo moves the whole dashboard between them, showing your role in each. With one workspace it is simply a label, so nobody gets a control they do not need.
"Answer every DM", the writing style, the human pause and short messages can each be set for one connected account instead of the whole workspace, so a shop's FAQ never answers DMs sent to a personal page. An account you have not touched simply follows the workspace setting.
A knowledge base can answer for every connected account or for exactly one. Use it when two of your accounts sell different things and share a workspace. A base you switch off keeps its documents and answers nothing.
Shipped 28 September 2026. Connect HookSend to Claude, Codex, Antigravity or any client that speaks the Model Context Protocol, and do the work by asking.
Describe the conversation you want and have the flow built, set up a campaign on your best reel, ask which post actually got the reach, or read your captured leads — from the AI client you already use. The connector offers 60 tools covering campaigns, flows, knowledge bases, the inbox, brand offers, Comment Guard, the storefront and your statistics.
You create a named token in Settings while signed in and it belongs to that one workspace, so nothing a caller sends can claim a different one. Only a fingerprint is stored, so the token is shown once and nobody at HookSend can read it back, and revoking it stops the client working immediately.
9 of the 60 tools refuse the first request and make your AI come back and ask you. Nothing gets deleted, nothing starts by itself, and buying a plan, changing billing or inviting a colleague all refuse an AI caller outright.
Every safeguard, in detailThe parts you never configure, and the reason a post that goes unexpectedly well does not cost you the DMs.
Every message sent and every comment hidden goes through Instagram's own API, paced inside Instagram's own limits. HookSend never asks for your password, never logs in as you, and never drives a browser pretending to be you, which is what the tools that get accounts restricted are actually doing.
Instagram allows 750 comment DMs an hour per account. Past that, HookSend queues each DM and sends it when its turn comes instead of throwing it away, so the thousandth person to comment on a reel that went unexpectedly well still gets their link. The send log shows what is queued and when it is due.
Every comment HookSend has seen is recorded, so a webhook Instagram redelivers, or a sweep that finds the same comment again, cannot send a second DM. Somebody who comments twice under one post gets one reply, not two.
Webhooks go missing occasionally, and a missed webhook is a customer who commented and got nothing. HookSend also sweeps your posts for comments it has not seen and processes those, sharing the same record so neither route can act on a comment twice.
What you pay, what you can add when a month runs hot, and how to reach us.
Every workspace starts with every feature switched on for 15 days, with no card: 50,000 DMs, 6 Instagram accounts, 500 AI credits, and 2 knowledge bases and 2 flows to try. When it ends you move to the Free plan, which does not expire, rather than losing the account.
Paid plans are monthly and never renew on their own, so nothing is taken from you without you doing something. Free carries 1,000 DMs a month on one account; the paid plans run from 7,000 on Starter up to 1,50,000 on Agency. Prices are in rupees in India and in dollars everywhere else.
Compare what each plan includesBuy 3,000 extra DMs at a time for {{dmpack}}. They are only drawn on once the month's own allowance is used up, soonest-expiring first, and a bought DM is spent once rather than coming back with the period.
On Business, Growth and Agency you can top up AI credits whenever you need more, on a slider. Pro includes 200 credits and cannot buy more, so a Pro workspace that runs dry upgrades instead. Credits bought on top last thirty days.
Beyond what your plan includes, one more knowledge base or one more flow costs {{slot}} and lasts 30 days. The number is how many can be live at once: save as many as you like and switch between them. Extras are only sold on a plan that already includes at least one.
Every order has an invoice you can open and print from the billing screen, with your workspace details on it. India pays by UPI; cards work everywhere else.
Share your referral link and earn between 10% and 30% of what each customer you bring pays, for 12 months per customer, renewals included. Your rate comes from how many paying customers you referred in the last twelve months, counted as a rolling window rather than an all-time total, so a badge earned this year is a badge you are currently earning. Take the money as a payout by UPI, or as credit against your own plan with no minimum.
Raise a ticket from inside your account, attach a screenshot, and follow the whole conversation there rather than losing it in email. Tickets are filed under eight categories including Instagram connection problems, billing, bugs and feature requests, each with its own number and status. Refund and data deletion requests have their own categories, and a deletion request is only accepted from inside the account, never by email.
The dashboard carries a meter for your AI credits and bars for the DMs, accounts and campaigns your plan allows, and a banner appears while there is still room to act rather than after sending has stopped. Every order has its own page, and the page you land on after paying verifies the payment with the provider rather than trusting the redirect.
A few things people ask about that Instagram's rules don't allow, or that HookSend was never built for.
Instagram doesn't tell any app when someone follows you, so no tool on the official API can message new followers automatically. HookSend greets people the moment they comment, reply to your story, mention you or message you.
HookSend doesn't publish or schedule posts, reels or stories. A campaign runs from the moment you switch it on until you pause it.
Instagram only lets a business message someone who has commented, replied to a story, mentioned the account or sent it a message first, so HookSend can't send a broadcast to all your followers.
Every workspace starts with a 15 day free trial of every feature, with no card, then moves to the Free plan. Paid plans are monthly and never renew on their own.
HookSend is built and run by HOOKSEND SOFTWARE, an approved Meta Tech Provider. Meta reviewed the app and approved every permission it uses. Every message and every hidden comment goes through Instagram’s official API, paced inside Instagram’s own limits. We never ask for your password, never log in as you, and never automate a browser pretending to be you, which is what the tools that do get accounts restricted are actually doing.
About HOOKSEND SOFTWARENo card. When the trial ends you keep a free plan that does not expire.
