What's new

Booking emails from your Gmail, a clear "not delivered" when a reply doesn't go out, and a filter for escalated chats still waiting on you.

Booking emails, sent from your Gmail

When someone books through ChatGenius, they can now get an email from your own address: a confirmation, a heads up if the time changes, a note if it's cancelled, and a reminder before it starts. Nobody has to dig back through DMs to find their time.

• Pick which ones go out. Each email has its own switch, and you choose when the reminder sends, from 1 to 48 hours before.
• Write them your way. Edit the subject and message, and drop in the customer's first name, the service, the date and the time.
• Turn it on in Calendar > Booking settings > Customer emails. Needs Gmail connected and the Creator plan or up. Once it's on, ChatGenius asks for an email when someone books.

Know when a reply didn't go out

If Instagram, Facebook, WhatsApp or SMS refuses a reply, the conversation now shows it as not delivered instead of sent, so you can follow up yourself. Any follow-up queued behind it is cancelled, and the AI no longer treats the failed message as something the customer saw.

Also new

• Needs reply. Escalated chats stay escalated until you resolve them, so answered ones pile up. The new Needs reply filter shows only the ones still waiting on a person, and the bar above Conversations counts them.
• A new row for every order in Google Sheets. Turn on New row for every capture in Connect > Google Sheets and a returning customer gets a fresh row each time instead of their old one being overwritten. Off by default.
• AI Configuration has a new look. Knowledge, Identity, Behavior, AI Gaps, Follow-ups and the rest are cleaner, with search and filters where you need them.
• Replies sent from Claude or ChatGPT now show the app's name in your inbox, and count as your own reply, so they end a flow that was waiting on you.
• One reply, not two. With Reply Speed set to Quick, Natural or Relaxed, a message the customer sends while a reply is waiting to go out is now answered in that same reply.

Fixed

• Follow-ups stop when the customer says no, says goodbye or tells you they booked. Before, a short "Nope" could still get another follow-up.
• Comment triggers no longer schedule a follow-up that Instagram was always going to block. If the person replies or taps a button, a follow-up is still scheduled.
• Long replies on Facebook are split into several messages instead of being rejected.
• When image analysis is off and someone sends a photo with a caption, the AI now answers the caption instead of a generic "thanks for sending that".
• The AI no longer swaps out normal replies that happen to use words like "booked" or "set for" when there's no booking involved.
• Phone numbers typed in Arabic or Persian digits are now captured as the lead's phone and reach your Google Sheet.
• A portal tab you're actively using no longer times out with a security error, and error messages are in plain words.
• Hiding or deleting a conversation removes it from the list without reloading the page.
• Follow-up times show in your timezone, and the 30-day insights range is right again after the clock change.

ChatGenius now works inside Claude on every plan, your flows can send email from your Gmail, and a week of fixes.

Run ChatGenius from Claude, on any plan

ChatGenius is now in Claude's Connectors directory. Connect it once and you can ask Claude about your account in plain words: "Who messaged me today and still needs a reply?", "Set up a comment trigger for the word GUIDE on my last post" or "Which posts got the most comments this month?"

• Every plan. This used to need the Webhook & API add-on. Now it's included on every plan, and Claude works within your plan's limits.
• One click to connect. Open ChatGenius in the Claude directory and click Connect, then approve it in ChatGenius. Works on the web, desktop and phone.
• ChatGPT too. It connects the same way once developer mode is on in ChatGPT. Steps are in Connect > AI assistants.

Connect in Claude

Flows can send email from your Gmail

The new Send Email step in the flow builder emails the contact from your own Gmail address. Use it to deliver the guide someone asked for, or confirm details once they share their email.

• Connect Gmail first in Connect > Gmail. ChatGenius can only send. It can't read your inbox, and replies land in your Gmail.
• Have a backup. If the contact has no email yet, the step takes its red path. Point it at a message so they still get what they asked for in the DM.
• Creator plan and up, 200 emails a day.

Also new

• See escalated conversations in one click. Conversations shows how many are escalated with a View button, and Escalated is now in the status filter.
• GoHighLevel gets your tags. Every tag on a ChatGenius contact now syncs to GHL, and tags you add inside GHL stay put. Comment trigger leads are tagged with the trigger's name, so you can tell which ad or post they came from.
• Your AI needs instructions before it goes live. Instagram and Facebook AI won't switch on until AI Configuration has at least a few lines about your business, so a customer never gets "I don't have that information" on day one.
• Upgrades are charged up front. Moving to a bigger plan mid-month now charges only the difference that day, and the new plan unlocks once it's paid. The amount you see before confirming is the amount charged.

Fixed

• The clock change on Nov 1 pushed some second follow-ups past Instagram's 24-hour window. Follow-up timing now counts real hours, so this won't happen again in March either.
• Turning follow-ups off now stops ones that were already scheduled.
• A comment trigger set for one post now wins over an all-posts trigger with the same keyword.
• An AI reply could be lost when our AI provider was slow to connect. It now retries quickly and still goes out.
• Messages with a job title, or button taps coming from ManyChat, no longer escalate a lead and pause the AI by mistake.
• AI comment replies and flows that start on what a customer means run on a newer, more reliable model.
• Knowledge gaps keep coming in when your list is full. The least useful old gap makes room, and more "I don't know" replies are caught.
• Lead capture no longer saves tracking numbers or numbers from links as phone numbers.

Links your AI sends are buttons now, plus AI working hours, faster Copilot review and two weeks of fixes.

Your AI's links are buttons now

When your AI shares a link from your instructions or knowledge base, it no longer pastes a long URL into the chat. The link goes out as a button under the message, labelled with what it does, like Book your call or Free course.

• Nothing to set up. Keep your links in your instructions or knowledge base like you do now. The AI writes the label.
• Works with Copilot. Drafts show the button, and approving sends it.
• Instagram and Facebook. WhatsApp, SMS and Telegram still get the plain link.
• Heads up if you reply from a computer. Instagram on desktop doesn't show buttons, so a message can look like it's missing its link. Your customer sees the button in the app. Please don't send the link again by hand.

AI working hours

Have your AI cover the hours your team doesn't. Pick the days and times it's on, and it switches itself on and off. Find it in Connect > Meta > AI working hours.

• Your timezone. Times are in your own timezone, and overnight shifts work.
• You stay in control. Turn the AI on or off by hand any time. The schedule picks up again at the next start or end time.

Also new

• Review Copilot drafts in one go. When drafts are waiting, Conversations shows how many with a Review button. It opens the oldest, and each approve, edit or dismiss takes you to the next.
• Customer language. Set the language your customers speak in AI Configuration > Identity. Flow defaults, booking times and other built-in messages use it until the AI knows which language a lead speaks.
• Longer AI instructions. Up to 20,000 characters, up from 15,000.
• Follow-ups read the conversation. They pick up where things stopped, never re-ask what the lead already answered, and stop when the lead says they booked.

Fixed

• The second follow-up was almost never sending. It now goes out, timed from the lead's last message, and respects your quiet hours.
• In flows, typing instead of tapping a button no longer picks the first button for the customer. If you reply yourself, the flow stops waiting for an answer.
• Comment auto-replies ignore instructions hidden inside a comment, so nobody can trick your AI into posting something you'd never say.
• The booking AI never mentions who has a taken time slot.
• Broadcasts tell you why they can't send (WhatsApp not connected, or no payment method) instead of failing quietly on resume or retry.
• Some incoming WhatsApp messages were saved twice in your inbox.
• We finished a full security review of ChatGenius this month and closed every finding.

Coming soon

We're already counting every tap on your AI's link buttons. Soon you'll see which links get tapped and who tapped them, the same way Link Clicks works on your comment triggers.

A free Link in Bio page, AI replies that wait like a person would, and a new Content Studio.

Your Link in Bio, inside ChatGenius

Claim your page at sumgenius.ai/@yourhandle and put it in your Instagram bio. It's free, and it's already in your sidebar under Link in Bio.

• Build it in one screen. Links, headers, text, videos and collections. Edits save as you type, and the phone preview updates as you go. Schedule a link or pin one to the top.
• Make it look like you. Start from a theme, then change the background, buttons, colours and font.
• See what gets clicked. Views, clicks, click rate, and where your visitors come from.
• Collect emails. Add a sign-up block and export your subscribers as a CSV.

Reply Speed

An answer that lands two seconds after a question can make people ask "are you a bot?". Now you can choose how long your AI waits before it replies. Find it in AI Configuration > Behavior.

• Instant, Quick, Natural or Relaxed. From right away up to 60 to 90 seconds after their last message. Instant stays the default.
• It still feels live. They see a typing indicator while it waits. If you jump in yourself, the AI's reply is dropped.
• Only AI replies wait. Flows, comment triggers and keyword replies still send right away.

Also new

• Content Studio was rebuilt. Calendar, Queue and Published views, a preview for each platform, and autosave. If a post fails on one platform, Retry only resends that one.
• Handoffs are quiet. When the AI passes a conversation to you, it no longer sends a canned "someone will be with you" line. You just pick it up.
• Broadcasts, WhatsApp Templates, Support, Community, Team, Contacts and Conversations all got the cleaner layout.

Fixed

• The AI only offers to book when you have an active service with a duration set. Calendar and AI Configuration tell you when nothing is bookable.
• Using ManyChat on the same account? Its button messages no longer pause your AI, and taps on its buttons no longer get an AI reply on top of ManyChat's.
• If another app already answered a comment privately, your trigger row now says so instead of showing zero DMs with no reason.

A new Comment Triggers page, a follower gate for your DMs, and Week and Day views in Calendar.

A new Comment Triggers page

Your triggers now sit in one list with the keywords, DMs sent, link clicks and last fired on every row. Click a row to see its details, or open the editor to see the DM exactly as it will look, sent from your own account.

• Test a comment. Type any comment and see whether your trigger would fire, before a real follower tries it. It's in each trigger's menu and in the editor.
• Only send this to followers. Turn it on for a button on Instagram. Followers get your reply right away. Everyone else is asked to follow first and gets an I followed button that checks again.
• Buttons are optional. A comment DM can now be plain text with no button. Triggers linked to a flow still need one.

Also new

• Calendar has Week and Day views, and times show in your timezone.
• Market Intel was rebuilt. Compare yourself side by side with any competitor, and send any post idea straight to Content Studio with Draft in Content Studio.
• Flows and Comment Manager got the same cleaner layout, with search and a channel filter in Flows.

Fixed

• A quick double click could send a team inbox reply twice or schedule a Content Studio post twice. It can't now.
• Content Studio stops you at 5 hashtags, the most Instagram and TikTok allow.
• Market Intel showed follower losses as gains on some days. Fixed.

See exactly who clicked your links, comment DMs that don't get lost, and a smarter AI Instruction Builder.

See who clicked your link

Link buttons used to show a number and nothing else. Now you can see the people behind it, so you know who to follow up with.

• Comment triggers. Click Link Clicks on any trigger card to see everyone who tapped the link in your DM, and when. This covers DMs sent from September 23 on.
• Flows. In a flow's analytics, click See who next to any link button's count. The list uses the same date range and channel filters as the panel, so the totals match.
• Export CSV. Both lists download as a spreadsheet, ready for a follow-up or your CRM.

Comment replies and DMs

• Failed sends keep retrying for up to 6 days. When Instagram or Facebook has a temporary problem, a comment reply or DM that fails gets saved and retried, the same message, until it goes through or Meta says it never can. Before, a few quick retries and it was gone.
• A failed DM no longer blocks that person. If a DM to a commenter didn't go through, they can still get one the next time they comment your keyword on that post.

AI Instruction Builder

• Now on Claude Sonnet 5.5. Open it with Describe Your AI in your AI settings (desktop).
• It asks before it writes. If your request is thin, it asks up to three questions first, and it already knows your services, FAQs and contact details.
• See every change, undo any of them. Each edit shows exactly what changed in your instructions, with an Undo button, plus suggested test messages to try it out.

Also new and fixed

• Content Studio post stats (views, reach, likes, saves) now fill in for Instagram and Facebook posts. They weren't being collected before, and that's fixed.
• If you use ChatGenius from Claude, the MCP now has 105 tools, including creating, editing and testing comment triggers.
• Some buttons in the portal no longer turn white on hover.

The portal is faster to move around, Market Intel opens in about a second, and comment triggers have folders.

Faster everywhere

We spent this week timing the portal instead of adding to it. Three things were making it feel slow and choppy, and all three are fixed.

• Switching sections no longer flashes empty. The page you are on stays put until the next one is ready, then it swaps in one frame. Your browser also stops re-downloading the portal's files on every click, which was most of the wait.
• Market Intel opens in about a second instead of twenty. Your Instagram insights are kept for six hours and refreshed in the background, and the Refresh button pulls live numbers whenever you want them.
• One slow page no longer freezes the rest. A slow Instagram or AI call used to block every other click in your tab until it finished. Fourteen of those endpoints now let the rest of the portal keep working while they finish.

Comment triggers

• Folders. Same as flows. New Folder in the header, a colour per folder, Move on any trigger. Deleting a folder puts its triggers back in Uncategorized and never touches the triggers.
• Your first trigger now catches the people who already commented. The moment a trigger built in setup goes live, we check that post for people who commented your keyword in the last 7 days and offer to DM them with one tap. Free accounts get this once, for the first 20 people.
• Rewind stops promising a DM it cannot send. Instagram allows one private reply per comment. When that reply was already used, we retried anyway and posted the public "check your DMs" first. Now the DM goes first and the public reply only follows once it lands.

Conversations

• Saved replies. If you answer DMs yourself, type / in an empty message box or tap the bolt next to it to pick a saved answer. It fills in the person's name for you and nothing sends until you hit send. Creator and up.
• Profile pictures stay. Instagram's picture links expire after a few days, so after a week the inbox showed initials for everyone. We now keep our own copy.

Also fixed

• The "Automation is off" tip came back on every page load after you dismissed it. It stays dismissed now, and comes back once if you turn automation off again later.
• On wide monitors, settings-style pages no longer stretch across the whole screen, and the Webhook page no longer leaves a gap next to the sidebar.
• The login page remembers how you signed in last time and marks that button.
• A Page or Instagram account can only be connected to one ChatGenius account at a time. A second attempt is refused and told which login already has it.
• If your Instagram connection ever drops, we get an alert the same day so we can reach out instead of waiting for you to notice.

Your account and billing moved inside the app, the portal has a light mode, and you can now connect Claude or ChatGPT straight to your ChatGenius account.

One app instead of two

Your settings, your plan and your agency accounts used to live in a separate dashboard outside ChatGenius. All of it is now under the account chip at the bottom of the sidebar. Old links redirect, so nothing you bookmarked breaks.

• Everything under your name. Profile, password and 2FA, language, plan and billing, agency accounts, help and sign out.
• There is a light mode. The moon and sun in that same menu.
• Stay signed in. Tick "Keep me signed in" and the portal stops throwing you back to the login screen every few hours. Add it to your phone's home screen and it opens on your inbox.
• Switch between monthly and annual yourself. Flip the billing toggle and your current plan offers the switch, with the price difference shown before you confirm.

Connect Claude or ChatGPT to your account

Ask an AI assistant what came in overnight, have it draft a reply in your voice, get it to build a comment trigger or a whole flow, or ask how you compare to the competitors you track. Go to Connect, then AI assistants, copy the address and paste it into your assistant.

• You choose how much it can do, read only or read and act, and you can disconnect it any time.
• Setup steps for Claude, ChatGPT, Cursor and Claude Code are on the page.

Part of the Webhook & API add-on, since it uses the same access layer as the REST API.

Conversations

• Rebuilt for phones. Threads open full screen, search and filters stay pinned, contact details and every action come up from the three dots. Desktop is unchanged.
• No more empty threads. Someone hearting, reading or editing a DM used to open a conversation with nothing in it.
• Story replies and mentions are saved now. When a trigger answered a story reply, the customer's message was never stored, so the thread showed your reply answering nothing.
• Profile pictures are back. About half the contact pictures we had stored were being cut short when we saved them, so they could never load. They fill back in as each person messages you.
• The Meta panel stops saying "Connected" when it is not, and the connect screen now tells you Instagram needs a Business or Creator account before you click.

Building on the API? It is roughly four times the size it was

122 endpoints and 31 webhook events, up from 32 and 16 a week ago.

• Build a flow from code. Ask for the node catalogue, send a flow, validate it and publish it. It opens in the builder laid out for you.
• Content Studio and booking: drafts, media, scheduling, publishing, availability and appointments.
• Analytics and Market Intel: flow performance, your account overview, competitor tracking, follower history, audience demographics and top posts.
• Set yourself up without asking us. Create and rotate API keys, and create, test and pause webhook endpoints, all through the API.
• Machine-readable spec at /api/v1/openapi.json; the reference is in the docs under REST API.

Finishing a flow got easier, the docs got a search box, and we now catch it when Instagram quietly stops your DMs.

Finish a branch without dragging a line

Adding the next step used to mean dragging a connection out into empty canvas and then finding something to drop there. Two easier ways in now.

• An unfinished branch shows itself. Any output with nothing wired to it goes hollow with a + through it, keeping its branch colour. You can see at a glance which branches still go nowhere. Click one and the node picker opens right there. Dragging still starts a connection exactly as before.
• Drop a step into a line you already drew. Click a connection and you get an insert and a delete on it. Insert puts a new node in between and rewires both sides for you, staying on the branch you clicked instead of jumping to the first one.
• New nodes stop landing on top of each other. A node added from a second branch no longer drops onto the one the first branch just made, and inserting into a tight pair pushes the rest of that path over instead of stacking on it.
• The palette search knows what nodes do, not just what they are called. Search escalate, hand off, tag, assign or custom field and you get the Action node. Search story, reel, ad, voice note, file or collab invite and you get the trigger that handles it. It used to only match the visible label.

The docs have a real search box now

Every guide is searchable, on the hub and on every article page, on your phone as well as your desktop. Results land on the exact section that answers the question rather than dumping you at the top of a long page.

• It answers as you type, with no round trip for the keyword half.
• You do not have to know our words for it. Type "my bot replies twice" and you get the page about duplicate messages, even though it shares no words with your question.
• Exact things still work as exact things, so message.postback finds the webhook reference.

We catch it when Instagram switches your DMs off

There is a toggle in the Instagram app, Allow access to messages, that switches itself off more often than it should. When it does, comments keep coming through and your automation still posts its public reply, but no DM can land. Nothing else looks wrong, so it can go unnoticed for days.

We already caught this on some connections. Now we catch it on all of them, and you get a banner in the portal and an email with the exact steps as soon as it happens.

If you want to check yours: Instagram app, Settings and activity, Messages and story replies, Message Requests, then under Connected tools turn on Allow access to messages. Our docs and help pages were pointing at an old menu name for this and now all say the same thing.

Building on the API? Your flow and consent data is now pullable

Flow performance and your opt-out list were visible in the portal and reachable nowhere else. Seven new read endpoints fix that.

• Flows and their lifetime stats, so a session or an event can be read against the flow it belongs to.
• Flow sessions and events. One person's path through a flow, or every path across a flow in a single feed you can poll incrementally.
• Opt-outs in one place. Meta and SMS merged into one list, with when and how each person opted out, and who has opted back in.
• Contacts now carry consent timestamps. First seen, last seen, and when that person first messaged you, which is the timestamp you would cite if anyone ever asks when they consented.

Read only, nothing renamed or removed, and the full reference is in the docs under REST API.

Also this week

• A broken link cannot get saved into a button any more. Someone pasted an Instagram share string, link plus caption, into a trigger button. It saved, and thirty people got a DM with a dead button. Every place you can type a URL now checks it the same way, and a stray space at the front gets trimmed instead of breaking the link.
• The typing bubble only shows when something is going to reply. With the AI off and no active flow on that channel, your customers no longer see "typing..." followed by silence.
• Landlines stop getting texted twice. When a carrier tells us a number cannot receive SMS, we remember it instead of trying again on the next send.
• Your billing month follows your timezone. If you signed up late in the evening somewhere west of us, your usage month was starting a day late. It now anchors to the day it actually was where you are.
• Anything a flow collected is cleared on the same 90 day schedule as conversations. An email or phone number is copied onto the contact record first, so you keep the lead and we stop keeping the raw copy.

Big week. Someone comments, and a whole flow starts.

A flow can start from a comment on its own

You no longer need to build a comment trigger and link a flow to it. Open a flow, set it to start from comments, type your keywords, and the flow itself handles the whole thing from the first comment onward.

• The first message is the private reply. Text, buttons, or an image card, sent as the one DM Instagram allows you to send a commenter who has never messaged you. One per person per post, so somebody commenting five times gets one reply.
• Scope it to one post or all of them. Same picker you know from comment triggers, and a scoped flow wins over a global one.
• Public reply variations so your public comment back is not the same sentence forty times under one reel.
• Clean Comments. Turn it on and the keyword comment is hidden once the DM actually lands. If the reply fails, the comment stays visible, because hiding it would leave that person with nothing.
• Rewind works on these flows. Build the flow today, scan a post from last week, and reach the people who already commented. It skips anyone your automations have already DMed on that post.
• The comment is available inside the flow. The post, what they wrote and the keyword they matched can all be used further down, including in HTTP nodes.

Your existing comment triggers are untouched and still take precedence, and flows attached to a trigger work exactly as before. This is a second way to build, not a replacement.

Edit a live flow without changing what is running

Editing an active flow used to change what your customers were walking through, the moment you saved. Now every flow keeps a draft next to its published version. You save as much as you want, and live traffic keeps running the published one until you press Publish. Anyone already partway through a flow finishes on the version they started.

• Discard draft puts the canvas back to what is live.
• Set live restarts a paused flow on its published version without shipping the draft you are still working on.
• Restore as draft pulls any saved version back onto the canvas to work from.
• Two tabs open no longer costs you work. If the flow changed somewhere else while you were editing, you are told and given the choice, instead of one save quietly erasing the other.

New integration: Google Sheets

The fourth integration, after GoHighLevel, Klaviyo and Google Calendar, and the first one built for people who do not run a CRM at all. Connect it and ChatGenius creates a spreadsheet in your own Drive with a Leads tab and a Bookings tab, then writes to it as the conversations happen. Settings, then Integrations, then Google Sheets.

• Every lead lands as a row. When it came in, which channel, name, username, email, phone, and a link straight back to the conversation it came from. Bookings get their own tab with the appointment time, service, status and confirmation code.
• The row grows with the conversation. Email now, phone number an hour later, a booking tomorrow that gets moved to Friday. It all lands on that person's row, so what you are looking at is where things actually stand.
• It arrives full. Connecting writes in the leads you have already captured, oldest first, so the sheet is worth opening the minute it exists.
• Leads and bookings are separate switches, if you only want one of the two tabs filling up.
• It is your spreadsheet. Sort it, filter it, add your own columns, chart it, share it with someone who has no ChatGenius login. ChatGenius only ever touches the one file it created and cannot see anything else in your Drive.

Also this week

• Branch ports you can actually tell apart. The success and error dots on an HTTP node sit further apart, each one is coloured, and every line takes the colour of the port it left. If you have an HTTP node, glance at it: a message wired to the error port instead of success ends the flow silently.
• The tester reaches every branch now. Simulate no response on a message with buttons and a timeout, type a wrong answer into a Collect Input node and watch the retry branch, and pick which button you want to start an attach-mode comment trigger from.
• Post pickers go back further than twenty posts. There is a Load older posts button in comment triggers, Rewind and the Comment Manager exclude list.
• Fallbacks for variables. Write {name|there} and it uses their name when you have it and "there" when you do not. Works in message text, button links and HTTP nodes.
• Two of your own connected accounts messaging each other will not loop. The conversation pauses and shows you why, and the messages are still saved.
• Building on the API? You can now send a private reply with buttons, get a webhook on every button tap, and see whether a commenter follows you.

Last week we said flow analytics was coming. It's here. Here's the week.

Flow analytics, on the canvas

Open any flow and switch to Analytics. Your flow is drawn exactly as you built it, with the real numbers on every node and every branch. No separate report to go find, no guessing which box the drop-off happened in.

• Every node shows its own story. How many people reached it, how many kept going, how many answered, how many stopped there and why, and how long it took them. A question node shows which reply people chose. A booking node shows what happened to the booking. A condition shows which way people went.
• Every branch shows its share. The line between two nodes carries the number of people who took it and what percentage that was, so the button nobody taps is obvious at a glance.
• Link clicks are tracked. Any URL button in a flow message now records its clicks, tied to the exact node and button. Repeat taps from the same person don't inflate it.
• Version history. Every time you save a real change, ChatGenius keeps a copy of that version of the flow. Analytics has a version picker so you can look back at how an older version actually performed, on its own canvas, without touching your live one.
• Filter by date range and channel, and compare against the previous period.

One thing worth knowing: the canvas only paints numbers from your current version of the flow. Runs from before your last edit are counted separately below it, because a node you've since changed or deleted can't honestly wear a number. Your lifetime totals in the flow list are unchanged.

The AI keeps its place in longer conversations

If you run a qualification funnel with four, five or six questions, the AI now holds on to which post or listing the conversation started from for the whole conversation, and reads far more of the original caption. The details it quotes back stay the right ones all the way to the last question.

Button taps carry through too. When someone taps a button in a comment trigger reply, your AI now sees which button they chose, so it moves on instead of asking again.

Rewind is steadier

Rewind now runs one job at a time per account and claims each person before it sends, so nobody on your list gets the same message twice.

It's also honest about where it stands. A job that stops partway through says so rather than sitting on a spinner, and your next scan picks up exactly the people it never reached.

Also this week

• Lead details are captured while you're handling the chat. If a customer gives their email or phone after you've taken over the conversation, it lands on their contact record the same way it would if the AI were replying.
• Copilot drafts read right from the second message on. The greeting meant for a first reply now stays on the first reply.
• Unreachable phone numbers are caught before we send, and your send log names the actual reason, so you can find the record in your CRM and fix it.

Your dashboard got rebuilt from the ground up, booking got faster, and Copilot got sturdier. Here's the week.

A new dashboard

The old dashboard showed everyone the same wall of numbers. The new one opens with the only two things that matter: is it working, and what needs you. Conversations waiting on a person and Copilot drafts waiting to send sit right under the headline, one click away.

Below that, everything is a module you can expand or collapse: Automation, AI Assistant, Delivery, Contacts, Appointments. The page opens the ones you actually use and quiets the rest, and whatever you collapse stays collapsed. Your layout is yours.

• It tells you things worth knowing. If one flow is doing most of your automation work, if a channel that was busy went quiet, or if your AI stopped replying while still switched on, the dashboard says so in plain words.
• Channel activity at a glance. Each connected channel shows its conversations and whether AI or automations are on, with activity bars in each channel's own color.
• Comment funnels show both stages. A flow started by a comment trigger reads "65 fired · 28 entered", so you can see your opt-in rate without opening a report.
• Your usage, honestly. Conversation and SMS usage each get their own bar, with your billing cycle dates on it.

Booking stopped asking for permission

When a customer has given the AI everything it needs, it books. No more "should I go ahead and book that?" after they already said yes. A simple booking that used to take five messages now takes two.

Reschedules and cancellations got the same treatment: moving an appointment never re-asks for details it already has, a question about cancelling never cancels anything, and before releasing a slot the AI offers to move the appointment instead. All of it works in every language your customers write in.

Copilot Mode is sturdier

If you run Copilot, drafts now behave everywhere: a waiting draft can't be sent twice, never leaks into analytics or your AI's memory as if it had been delivered, and old drafts resolve instead of sitting forever.

Waiting drafts also show on the new dashboard, in the AI Assistant card and right under the headline, so a customer is never quietly waiting on a reply you didn't know existed.

Also this week

• Appointments now live on the contact. Open a contact and see their bookings; open a booking and see who it belongs to.
• Comment Triggers rebuilt for your phone. The whole section now works properly on mobile.
• Messages over your plan limit are kept. If you hit your conversation cap, incoming messages are stored and waiting for you. Nothing is lost.
• SMS opt-outs are honored on every path, and we now record what the carrier actually delivered, not just what we handed off.
• The AI no longer claims it can't see images. It can, and now it says so.

Coming Soon

Flow analytics. As of this week your flows record every step a customer takes through them, and we're building the reports on top: where people drop off, which paths convert, and which button nobody taps.

Your contacts can now hold whatever matters to your business, your flows fill it in for you, and replies you send from the Instagram and Facebook apps finally show up here.

Custom Fields

Tags tell you whether something is true about a contact. Custom Fields tell you what: budget, renewal date, party size, referral source. Up to 15 fields, as text, number, date, or yes/no.

They show up in Contacts, on the conversation panel while you're reading a thread, and as columns in your CSV export. Every plan, Free included, just like tags.

Your flows fill them in

Collect Input has a new option: also save this answer to a contact field. Ask someone their budget once and it's on their record for good. A new Set Custom Field action stamps a value silently, without asking anything, which is handy for recording intent from a comment flow.

The AI flow builder knows your fields, so describe what you want and it wires the saving up for you.

Replies from the Instagram or Facebook app now show up here

If you answered a customer from the Instagram app, Messenger, your Page inbox, or Business Suite, ChatGenius never saw it, and your AI could reply on top of a conversation you were already handling.

Those replies now land in your inbox labelled "Sent from app", and automation stands down on that thread for five minutes so you can finish. Nothing to switch on, and these are never billed as conversations.

Faster replies, and clocks that agree

Your AI moved to a faster model on the main conversation path. Greetings now come back in about a second instead of two and a half, and language detection is more accurate.

We also fixed a clock mismatch that was shrinking Meta's 24 hour reply window to roughly 20, so flows gave up on customers four hours early. Inbox times now show in your timezone, and your AI reads each appointment in that appointment's own timezone.

Webhook & API: the rebuilt dashboard is live

Your existing setup carried over untouched. You now get up to three endpoints, each with its own signing secret, event selection, pause switch, and delivery logs, plus named API keys you can rotate or revoke one at a time.

The docs went from five pages to ten, with a full REST reference, every payload for all 15 events, and a Markdown version of each page. Custom Fields are readable and writable through the API, with a new contact.field_changed event.

Fixes and polish

• Copilot Mode can no longer be bypassed. Booking replies and queued follow-ups were still going out unapproved. Every path checks now.
• You'll see it when a plan is ending. Schedule a cancellation and your dashboard shows the date it runs until, with a one-click way to keep it.
• US numbers without a country code. On SMS, a plain 10 digit number was read as international and quietly never arrived. Now resolves to +1.

Want to see everything we ship? Our What's New page keeps a running list. Questions or feedback? Just reply to this email, or book a call.

This week your AI learned 14 new languages, flows learned to run exactly where you want them, and a big upgrade is on the way for everyone on the Webhook & API add-on.

Your AI now speaks 28 languages

We doubled language support from 14 to 28 languages. Your AI detects the language a customer writes in and answers in it automatically, including right-to-left languages like Arabic and Hebrew, and newly added Greek, Thai, and Ukrainian. Nothing to configure.

If you've written a rule like "always reply in Spanish" in your AI Instructions, that rule wins: detection never overrides what you've told your AI. And if you sell in euros, pounds, or anything other than dollars, you can now set a currency on each service, and your AI quotes prices with the right symbol and formatting.

Flows: choose exactly which channels each flow runs on

A flow can now target any combination of Facebook, Instagram, and WhatsApp instead of one channel or all of them. Want a promo that only runs in Instagram DMs while your WhatsApp stays strictly support? Tick the boxes you want and you're done.

The builder also got smarter about impossible setups: if a trigger can't reach a channel, that channel is greyed out with the reason right there, and each flow card shows badges for where it runs. You can't accidentally build an automation that never fires.

For Webhook & API subscribers: new events now, a new home soon

If you're on the Webhook & API add-on, there's a lot that's live right now: new webhook events for appointments booked, cancelled, or rescheduled, leads captured (email and phone as separate events), conversations escalated to a human, and failed sends. Event payloads now include the contact's tags, stage, and assigned team member, Telegram works through the API end to end, and a full REST API lets you read and update contacts, conversations, tags, and appointments from your own systems.

And coming next: a completely rebuilt Webhook & API page. Multiple webhook endpoints, each with its own signing secret and event selection, multiple named API keys you can rotate or revoke instantly, and delivery logs and health tracking per endpoint. It's built and in final testing now, and fully refreshed developer docs land with it. Nothing changes for your current setup, and we'll let you know the moment it's live. Until then, if you want to wire up any of the new events or API, just reply and we'll get you set up.

Fixes and polish

• Customers see the typing dots. While your AI prepares a reply on Facebook and Instagram, the customer now sees the typing indicator, so it feels like a real conversation instead of a silent pause.
• SMS and Telegram freed from Meta's clock. Meta's 7-day messaging window only applies to Meta channels, but SMS and Telegram conversations were showing "window expired" warnings and blocking claims. Those channels now work without the false alarms.
• Telegram, smoothed out. Sending from your dashboard now works on Telegram conversations, and messages a customer sends after a team member claims the chat land in your inbox instantly instead of getting lost.
• Opt-outs respected on every channel. A customer typing STOP now works on Telegram too, and anyone who opts out on any channel is reliably excluded from follow-ups everywhere.
• No more page reloads in your inbox. Switching conversation filters now updates the list in place, keeping your scroll position and making the whole inbox feel faster.

Want to see everything we ship? Our What's New page keeps a running list. Questions or feedback? Just reply to this email, or book a call.

This week building an automation is as simple as filling out a form, every customer gets a permanent home in the new Contacts section, and you can choose to keep answering after you hit your monthly limit.

Build an automation in plain English

Creating a flow now starts with a choice: Simple reply, a template, a blank canvas, or generate with AI. Pick Simple reply and you get a plain form: type the keyword or event that triggers it, write your reply, add media and up to three buttons, and watch a live phone preview update as you type. No canvas, no connecting boxes.

Behind the scenes it is a real flow, so if it ever outgrows the form, one click on Advanced mode opens it in the full builder with nothing lost. And with this, Quick Replies and Flows are now one system: anything a quick reply could do, a simple flow does, all in one place. If you had quick replies, we already converted them into flows for you. Nothing for you to do.

Meet Contacts: every customer, one place

The new Contacts section gives every person who messages you a permanent record: name, email, phone, tags, lead score, and a hot/warm/cold temperature, kept independently of the conversation itself. Clear out old conversations and the customer record stays. Search and filter your whole list, export it to CSV, and edit any detail by hand knowing incoming platform data will never overwrite your corrections.

Optional: keep answering after you hit your monthly limit

Until now, hitting your plan's monthly conversation limit meant your AI simply stopped replying until the next cycle. If you would rather keep the conversations going, you can now switch on Conversation overage and pay $0.03 per extra conversation instead of going quiet. It covers Instagram, Facebook, WhatsApp, and Telegram.

It is off unless you turn it on, and only the account owner can. Leave it alone and nothing changes: your limit works exactly as it does today, and you will never be charged a cent past your plan. If you do switch it on, your plan screen shows what you have used and what you have accrued as it happens, and anything extra lands on your next invoice. Available on paid plans.

Your Instagram connection now renews itself

Meta issues Instagram connections that expire every 60 days. ChatGenius now renews yours automatically about two weeks before that deadline, quietly, in the background. Your account stays connected without you ever having to log back in or reconnect. Nothing to turn on.

Fixes and polish

• One home for your automations. Comment Triggers and Flows now live together under a single Automations group in the sidebar, so everything that replies for you is in one place.
• Rename a flow on the spot. Click a flow's name in the list to rename it right there, no need to open the builder. And the "Updated" time on each flow now shows in your timezone.
• Facebook comment buttons stay put. Buttons on Facebook comment-to-DM messages used to disappear the moment someone typed instead of tapping. They now stay pinned to the message, so your customer can always come back and tap one.
• Phone numbers from anywhere in the world. Lead capture now recognizes international phone formats, so a customer's number lands in your Contacts correctly with the right country code, wherever they are.
• Buttons show in your inbox too. Buttons your automations send now appear in your inbox exactly as your customer saw them, so you always know what they were offered.
• You'll always know when automation is off. If you have AI or automation switched off, those pages now clearly say so, instead of letting you build something that is not set to run yet.

Want to see everything we ship? Our What's New page keeps a running list. Questions or feedback? Just reply to this email, or book a call.

This week you can connect the WhatsApp number you already use on your phone, greet ad traffic the second it lands, and your AI now understands voice notes on Messenger and Instagram.

Connect the WhatsApp number you already use

Until now, connecting WhatsApp meant using a number that was not already on the WhatsApp Business app. That ruled out most businesses, because the number you actually talk to customers on is the one already in your pocket.

Now you can connect that number. Keep using the WhatsApp Business app on your phone exactly as you do today, and ChatGenius works the same conversations alongside you. Look for Connect your existing WhatsApp number in Settings. Available on Professional and above.

New trigger: Ad click

If you run ads that open a chat, you can now fire a flow or a quick reply the moment someone arrives from your ad. No keywords, no guessing what a new lead might type first. It works even if all they say is "hi." Facebook, Instagram, and WhatsApp ads are all covered.

Running more than one ad? You can scope a flow to a specific ad, so each campaign sends its traffic into its own funnel. Leave it blank and the flow greets everyone who arrives from any ad.

Your AI now understands voice notes

When a customer sends a voice note on Messenger or Instagram, your AI listens to it, understands it, and answers like it would to any other message. Voice notes also feed your funnels and automations, so a spoken question can start a flow just like a typed one.

Funnels that hold up in the real world

Real customers do not follow the happy path. They tap a button a day later, they tap it twice, or they type the answer instead of tapping. Your funnels now handle all three.

• Late taps still work. Someone who comes back a day later and taps a button picks up right where they left off.
• Double taps stay clean. An impatient customer tapping twice no longer gets duplicate messages or a jumbled thread.
• Typed answers count. If someone types "Online" instead of tapping the Online button, your funnel keeps going instead of stalling.

Comment triggers re-engage on every new post

If you run a comment trigger across all your posts, a repeat commenter now gets your DM again on each new post they comment on, while still being capped at one DM per person per post. Your best commenters are the ones who come back, and now every post you publish is a fresh chance to reach them.

Fixes and polish

• Exact match keywords. Keyword triggers now have an exact option, so a keyword like "free" fires only when someone sends exactly that, not inside "when are you free."
• Photos and videos appear instantly in your inbox. Media a customer sends now shows up live in an open conversation, with no refresh.
• Smarter handoffs to your team. Escalation is more precise, so a shared reel or an everyday turn of phrase like "horrible weather" no longer pulls a human in for no reason.
• You stay signed in. The portal keeps your session alive while you work, so a longer form does not lose what you typed.
• Cleaner calendar. Long event titles from a synced calendar now fit the month view instead of stretching it.
• Easier document import. Importing a knowledge document now opens in a proper window, wherever you click it from.

Updated Terms and Privacy Policy

We refreshed our Terms of Service and Privacy Policy. They now spell out exactly when we can and cannot access conversation content, list every vendor we use, and set a 72 hour breach notification commitment. You will see a short prompt to accept them next time you sign in.

Want to see everything we ship? Our What's New page keeps a running list. Questions or feedback? Just reply to this email, or book a call.

This week your AI gets a built-in safety layer that keeps every answer true to your setup, a new way to hand your lead magnet straight to customers in the DM, plus more room to build ahead.

Every reply is checked against your setup before it sends

Your AI now has a built-in safety layer. Before any reply goes out, it checks the answer against your own setup, your instructions, knowledge base, and services, so what your customers hear always matches what you actually offer. It runs on every plan, automatically, with nothing to turn on.

Deliver your lead magnet right in the DM

You can now attach a PDF to a quick reply, a flow, or a comment trigger, and your customer gets the actual file in the chat instead of just a link. It is the easy way to hand over a guide, menu, price sheet, or freebie the moment someone asks. Works on Instagram, Messenger, and WhatsApp, up to 25MB.

Plan ahead without hitting your limits

Your plan's limits on flows, comment triggers, and quick replies now count only what is live. Draft and stage as many as you want, and your limit only applies when you switch one on. Build ahead, keep them paused, and go live when you are ready.

A sharper, more natural AI

We upgraded the model behind your replies. It follows your instructions more closely, keeps to your funnel, and sounds more natural in conversation.

Fixes and polish

• Facebook Pages in Business settings now connect. If your Page lives in a Meta Business Portfolio, the Connect with Facebook step now finds and connects it like any other.
• Stickers and reactions no longer interrupt a flow. A sticker, reaction, or collab invite in the middle of a flow no longer sends a stray "tap one of the options" message.
• Export a flow, keep its comment trigger. Exporting a flow now brings its comment trigger with it and rebuilds it in the account you import into, so a comment funnel moves in one step.
• Your AI on/off setting always sticks. Your on/off toggle is now the final word, no matter what else you change in your settings.

Want to see everything we ship? Our What's New page keeps a running list. Questions or feedback? Just reply to this email, or book a call.

This week your dashboard learned three new languages, your AI got smarter about the calendars and corrections you feed it, and your premium analytics are reading clean again.

Your dashboard now speaks your language

The entire ChatGenius portal is now available in Portuguese, Spanish, and French, alongside English. Pick your language in Settings and every screen translates instantly, and a first-time login already greets you in your browser's language. Your customers' names and messages are never touched, only the interface translates, so your conversation data stays exactly as it came in.

Two-way GoHighLevel calendar sync

If you book through GoHighLevel, your AI can now see what is already on that calendar, not just write to it. Appointments booked directly in GHL now count as busy time, so the AI stops double-booking over them, the same way Google Calendar already works. One note: if you connected GHL a while ago, reconnect once so it can read your calendar.

Teaching your AI now fixes the right line

When you resolve an AI Gap and feed the answer back into your instructions, ChatGenius no longer just tacks it on at the end. It now finds the line that actually needs updating and rewrites just that one, so a new answer like "the trial is 14 days" replaces the old "7-day trial" instead of leaving both to contradict each other. You see a clear before-and-after preview, can edit it, and nothing is saved until you approve it. Pure facts can be filed straight into your Knowledge Base instead. Your AI gets cleaner and more consistent every time you teach it.

Deep Insights analytics, rebuilt and accurate

Instagram quietly renamed and retired several of the data points behind Market Intel's Deep Insights over the past year. We rebuilt that whole section against Instagram's current data, so your views, reach, post performance, engagement rate, and audience demographics (age, gender, top cities) all report correctly again. Post covers show their real image too. This is a Premium feature, and it is now reading clean on live accounts.

Fixes and polish

• No more silent replies. We fixed a case where the AI could occasionally come back empty and not respond at all. Those messages now go through reliably.
• Tighter calendar protection. We closed two edge cases where a very full Google or GoHighLevel calendar could still let a slot slip through and double-book. Busy times now block correctly even on packed calendars.
• Walk-in questions answered properly. If you have booking turned off, the AI now handles "can I just drop in?" style questions cleanly instead of pointing at a booking flow that is not there.
• A heads-up before you run out. On paid plans, you will now get an email when you hit your monthly conversation limit, so a busy month never goes quiet without warning.

Want to see everything we ship? Our new What's New page keeps a running list. Questions or feedback? Just reply to this email, or book a call.

Quick one, good news. We just raised the monthly conversation limits on every paid ChatGenius plan. Same prices, a lot more room.

New conversation limits

• Creator ($29/mo): 500 → 1,500 conversations/month
• Professional ($99/mo): 2,500 → 8,000 conversations/month
• Business ($297/mo): 10,000 → 25,000 conversations/month

Nothing for you to do. The new limit is already active on your account and your price stays exactly the same.

Why the bump? Plans were overdue for it, and our setup can handle far more than the old caps allowed. We would rather you have room to grow than hit a ceiling on a good month.

On the Free plan? You still get 50 conversations a month to test everything end to end. If you have been bumping against that, Creator now gives you 1,500 for $29.

Thanks for building with us. Reply anytime, I read these.

Jared
Co-founder, SumGeniusAI

This week your AI can show, not just tell. It can send a photo of what a customer is asking about, you get finer control over what each teammate sees, and a handful of fixes make conversations come through cleaner.

Your AI can send a photo, not just describe it

You can now attach an image to any service in your Knowledge Base, a product shot, a menu, a before-and-after, a room. When a customer asks about that service, the AI sends the photo right in the conversation alongside its answer, so people see what they are buying instead of just reading about it.

You upload it once on the service, with an instant preview, and it shows as a thumbnail on the service card so you always know which ones have a photo. The AI decides when an image genuinely helps and leads with it, then gives the full answer in text. It works on Instagram, Facebook, and WhatsApp, on every plan that has services, with nothing to turn on.

Choose which channels each teammate can see

If you have team members, you can now pick exactly which channels each person works in: Facebook, Instagram, WhatsApp, SMS, or Telegram. Set it on the invite or right on your team roster. Leave it alone and everyone keeps full access, so nothing changes until you decide to narrow it. A teammate limited to Instagram and Facebook will not see your SMS or WhatsApp conversations anywhere, the inbox, the dashboard, the metrics, or exports. Changes take effect on their next page load, no re-login needed.

Fixes and polish

• No more double replies. When a customer fired off two quick messages back to back, the AI could occasionally answer both with two near-identical replies in a row. It now waits a moment for the full thought before responding, so conversations read cleaner on every channel.
• WhatsApp QR codes stay working. Saved codes used to break and show a placeholder after a while. We now generate them ourselves from your permanent WhatsApp link instead of relying on a Meta image that could expire, so codes stay sharp for print and every existing one is already fixed.
• Copilot now sends images too. If you use Copilot Mode, approving a draft that includes a photo now delivers the image, not just the text. Pictures from your Knowledge Base and your new service images both come through.
• Fixed: long service descriptions getting cut off. We caught a bug that was trimming longer service descriptions before they reached the AI, so a few answers could trail off mid-detail. The full description now comes through, and those answers are complete again.

This week is about trust and context. Your AI can now draft every reply for you to approve before it sends, show you the questions it could not answer, and treat comment-to-DM leads like it knows the post they came from.

Copilot Mode: approve every AI reply before it sends

Not ready to let the AI talk to your customers on its own? Now you do not have to choose between AI off and fully automatic. Flip on Copilot Mode in AI Config, and every AI reply becomes a pending draft in your Conversations view. Nothing reaches a customer until you tap Approve and Send, and you can edit it first or dismiss it entirely.

It is the safe way to watch your AI handle real conversations before you trust it with the keyboard. Drafts show up live, approving sends as your bot through the same checks as a normal reply, and a newer draft replaces an old one if the customer messages again so you are never approving something stale.

Everything you built by hand keeps running at full speed. Quick replies, comment triggers, flows, and your own team replies all send immediately as before. Copilot only holds the AI's own generated replies. It is off by default, so nothing changes until you turn it on, and you can take the training wheels off whenever you are ready.

AI Gaps: see what your AI could not answer, and teach it in one click

Every AI eventually hits a question it cannot answer well. Until now you had no way to know which ones. The new AI Gaps tab (AI Config → Build → AI Gaps) collects those moments, the question your customer actually asked, cleaned up into a clear version, and what the AI said. Many cards come with a suggested answer already drafted from your existing knowledge.

Hit Teach your AI and the answer goes into your Knowledge Base immediately, so the very next customer who asks gets it right. Repeat questions stack into a count ("Asked 7 times") so you can fix what matters first. Detection and viewing are on every plan, including Free, and your AI gets measurably smarter the longer your account is live.

Smarter comment-to-DM follow-ups

When someone comments your keyword and the AI follows up, it now knows the post they came from. So when they reply "ok how much is it?", the AI answers about the offer in that post, using your own caption, instead of treating them like a stranger who appeared out of nowhere. This is the highest-intent moment in your whole funnel, and it now has context. It works automatically on every existing comment trigger, on every plan, with nothing to turn on.

Conversation Starters

Give people something to tap when they open your Instagram or Facebook DMs. The new Starters tab (AI Config → Build → Starters) lets you set up to four questions that show before anyone types, plus a Get Started button on Facebook. A tap sends the question as the customer's first message and the AI answers it, so the awkward blank-screen moment turns into a real conversation. Save and Publish pushes it straight to your DM window. All plans, including Free.

Fixes and polish

• We now catch when Instagram quietly turns off DM access. There is an Instagram setting that can switch off message access without disconnecting your account, and when it is off your AI silently stops receiving DMs. If that happens, you will now get an email and a clear banner with the exact toggle to flip back on, instead of wondering why things went quiet.
• More reliable AI replies on messy text. We fixed a case where messages with broken emoji or garbled characters could cause the AI to fail to respond. Those now go through cleanly.

This week was about getting started faster, managing more than one account, and a handful of additions for advertisers and global audiences.

A faster, guided start

New accounts no longer land on a blank dashboard. When you first sign in, ChatGenius asks one question, what do you want to set up first, and takes you straight there: comment-to-DM, AI replies, appointment booking, or content. If you have not connected a page yet, it points you to that first.

If you ever step away mid-setup, you will now see a clear Resume setup banner that drops you right back where you left off, instead of a dead end. And when you open comment triggers for the first time, Browse Templates is front and center so you can start from a working example rather than a blank form.

Manage more than one account

If you run ChatGenius for more than one brand or for clients, the new Agency mode lets you switch between accounts from a single login, with billing controls limited to the account owner. If this is something you would use, just reply and we will get you set up.

Also shipped

• WhatsApp broadcasts now show when each person read your message. Open any campaign report and you will see a per-recipient read time, not just an overall percentage.
• Ad attribution into your funnels. If you run click-to-WhatsApp ads, you can now pass the ad click ID into flow links, buttons, and checkout URLs (for example a Stripe link), so your conversions match back to the ad that drove them.
• Czech is now supported. The AI now detects and replies in Czech, our 14th language, across Instagram, Facebook, and WhatsApp.

Fixes and polish

• Cleaner on mobile. The setup and template pop-ups now fit your screen properly on phones, with nothing clipped behind the top bar or bottom menu.
• More reliable billing and limits. We tightened how plan limits and features apply when an account changes plans, so what you have access to always matches the plan you are on.

This week we focused on speed, clarity, and a few fixes worth knowing about, plus a more generous free plan.

A faster, clearer portal

We reworked how the dashboard updates. Creating, editing, or deleting a comment trigger, quick reply, or knowledge base entry now happens instantly in place, your counts update live, and you keep your place on the page. No reloads, no waiting.

Your Overview and activity views are now fully live too. Metrics and status update in real time, and every list, Conversations, Leads, Knowledge Base, Analytics, and Follow-ups, points you to the next action instead of showing a blank table. New accounts land on a focused setup view that makes the first step obvious.

A few fixes worth knowing about

• Connection errors now explain themselves. If a Facebook or Instagram connection does not go through, you get a clear message and the exact next step, right on the settings page.
• Escalations now hold the AI even when no one is online. Once a chat is handed to a human, the AI steps back and waits, so an escalated customer never gets an automated reply on top of the one you are about to send.
• Comment-to-DM is more reliable at scale. We rebuilt how comment triggers track who has already been messaged. If one of your posts gets a flood of comments, the same person will not accidentally get two DMs, and your automated replies stay fast no matter how busy a post gets.

Also shipped

• Inline explanations for every metric. A quick tap on any label across Analytics, Calendar, Flows, and Quick Replies tells you exactly what the number measures.
• Refined confirmations. On-brand confirmation dialogs throughout the portal, and one-click conversation claiming with a quick undo.
• A new Chrome extension for Creator and up: captions, hashtags, and post ideas in your browser, powered by the same AI as Content Studio.

A more generous free plan

We raised the free tier to 50 conversations a month, and kept the pricing model that sets us apart: you pay per active conversation in flat tiers, so growing your audience never raises your bill. If you know someone weighing their options, the free plan is now more than enough to put ChatGenius through its paces.

More documentation

We added a batch of guides at sumgenius.ai/docs: AI Configuration, Knowledge Base, Quick Replies, the AI Instruction Builder, Multilingual replies, Team Inbox & Roles, Escalation & Human Handoff, Appointment Booking, Leads, Connect WhatsApp, Content Studio, and Broadcasts & Templates.

If there is something you would like documented, just reply and let us know.

A big week for comment-to-DM. You can now reach people who already commented, keep your comment section clean, and see exactly which ad started a conversation.

Rewind: DM people who already commented

A Reel pops off after the fact, or comments slip through during a Meta hiccup, and a batch of warm leads commented your keyword but never got the DM. Rewind goes back and fixes that. Pick a post, scan its existing comments, and back-fill the comment-to-DM to everyone who commented before your trigger was catching them.

How it works

1. Open a comment trigger and pick a post to Scan
2. See a preview before anything sends, e.g. "142 comments · 38 matched · 31 will be DMed"
3. Confirm, and it sends gently in the background within Meta's rate limits

Guardrails are built in. It only DMs people who commented in the last 7 days (Meta's private-reply window), never double-DMs anyone your trigger already reached, and it mirrors that trigger's exact setup, with a per-scan option to skip the public reply on big back-fills. Available on Creator and up.

Clean Comments: hide the keyword comment after the DM sends

The classic complaint about comment-to-DM is that it fills your comment section with "GUIDE", "SHOP", "BLAST" spam. Flip on Clean Comments for any trigger and the comment that fired the DM gets hidden right after, so the public view stays tidy while you keep the lead. It's hide, not delete: the comment still counts as engagement to Meta and stays visible to its own author, so nobody feels censored, and it's fully reversible. Works on live triggers and on Rewind back-fills. Opt-in, off by default, on every tier.

See which ad started each Messenger & Instagram conversation

If you run click-to-message ads, ChatGenius now captures which ad drove each Messenger or Instagram DM, the same attribution that's been live for click-to-WhatsApp. New conversations get a Messenger Ad / Instagram Ad badge, a CTM Ads filter in your inbox, and an Ad Attribution card showing the ad title, ad ID, and the custom ref you set in Ads Manager. That ref is what lets you tie a conversation back to a specific ad for cost-per-lead math, no native Meta reporting or extra tracking tool needed.

Also shipped

• Duplicate a comment trigger. One click copies a trigger's keywords, DM and public-reply content, buttons, and AI config. The copy starts paused and named "(Copy)" with fresh stats, so you can tweak the one thing that differs and enable. Great for running the same setup across multiple posts. All tiers, within your trigger limit.
• Quick Reply "send only once per customer." A per-reply toggle so a welcome or promo reply fires at most once per person. Returning customers who reuse the keyword fall through to the AI instead of re-triggering it. Opt-in; your FAQ-style keywords keep answering every time.
• Escalation brief now shows in your portal. When a chat escalates, the AI's short "what they want, what's been tried, why they're upset" summary now appears at the top of the conversation in both the client and team views, not just the email. Whoever picks it up has the context without scrolling the thread.

New: a help docs site (work in progress)

We're building out proper documentation at sumgenius.ai/docs, step-by-step guides for setting up and getting the most out of ChatGenius. It's early and we're adding to it every week, but these are already live:

• What is ChatGenius, the 2-minute overview
• Connect Instagram & Facebook
• Your First Automation, your first comment-to-DM or quick reply
• Comment Triggers & Rewind, everything in this week's email, in depth
• Visual Flow Builder
• Webhook & Send API, for developers

More guides are on the way. If there's something you'd like documented, just reply and tell us.

Big week if you take bookings. The AI is now end-to-end smarter about virtual appointments — plus four upgrades that quietly cut steps out of your day.

Virtual bookings now ship with a Google Meet link

If you run virtual sessions — coaching, consults, demos, discovery calls — ChatGenius can now create a Google Meet link for every booking and email the calendar invite straight to your customer. No copy-pasting links, no follow-up "here's the Meet" DM, no surprises when the invite hits their inbox.

How to turn it on

1. Connect Google Calendar if you haven't (Settings → Calendar)
2. Flip on Send Google Meet invites for virtual services
3. In AI Config → Service Menu, tick Virtual meeting on whichever services need a Meet link

The AI booking agent now also knows which services are virtual — so it'll naturally weave "I'll send the Meet link to your email" into the conversation while collecting the customer's address. In-person services are completely unaffected; they keep booking exactly as before.

AI bookings now mirror into GoHighLevel

If you run on GHL, pick a calendar in Settings → GoHighLevel and every AI booking, reschedule, and cancellation will sync into it — alongside the contact sync you already had. Show/no-show stats, workflows, and reminders all live where you already work. Existing GHL connections need to reconnect once to pick up calendar permissions.

Quick Replies now fire on photos, voice notes, videos, and files

Four new triggers on the Quick Reply dropdown — Image Received, Voice Note Received, Video Received, and File Received. The reply fires the second that attachment lands, no caption needed. Great for "Got your photo, team will take a look" acks, prescription / insurance card uploads, quote routing on product shots, or a quick acknowledgement on voice notes before a human jumps in.

Build Comment Triggers without leaving the Flow Builder

When a flow uses a Comment Trigger as its source, you can now create or edit the trigger right inside the builder. Same form, same fields, same buttons — just inline. No jumping to another tab, no losing your unsaved flow halfway through setup.

The AI stopped promising a human when nobody's online

Before, the AI might say "let me grab someone for you" at 3am with no one around. Now it checks whether any team member is actually online and adjusts the wording — either escalates to a real agent, or flags the conversation for follow-up ("the team will get back to you"). No more contradiction between the AI's promise and what actually happens.

Four upgrades this week that change how you set up, attribute, and respond.

1. Claude now writes your AI instructions for you

Describe your business in plain English. Watch Claude build your custom instructions live. Test every change in a side-by-side bot before saving. No more blank textarea. Powered by Claude Sonnet 4.6, available on every tier.

2. WhatsApp QR Codes with source tracking built in

Print scannable codes for signage, packaging, bio links. Every scan auto-tags the conversation so you finally know which QR drove the lead — no UTM hacks, no third-party redirects. Pro and above.

3. Comment triggers can now AI-generate the DM

Flip on AI for any trigger, write a one-line goal, and every commenter gets a personalized DM using your knowledge base and brand tone — instead of one of 5 canned templates. Available on every tier.

4. Image, video, and PDF headers on WhatsApp templates

Hero banners on promos, carrier logos on shipping updates, invoices delivered as the actual PDF. Way more clickable than text-only. Pro and above.

Also shipped

• Independent AI + Automation toggles for WhatsApp, SMS, and Telegram — mix-and-match per channel.
• Keyword automation fires in ~2 seconds, not 5+. No more lag on "help" or "menu" replies.
• AI calmer on casual chat — thank-yous and shared reels get a warm acknowledgement and stop, no reflexive product pitch.
• Fraud auto-flag — impersonator reports and gift-card scams pause for human review with a Fraud badge.

Four upgrades this week that change how your AI greets, tags, and delegates

1. Welcome Message — a guaranteed first-impression greeting

Set a custom greeting in AI Config → Identity and ChatGenius will send it verbatim on the first message of every new conversation — bypasses the AI entirely, 100% reliable, works in any language. Pick Combined to prepend it to the AI's first reply, or Standalone to send the greeting alone and let the AI handle the customer's next message (saves tokens and feels more human). Built for clients who want a consistent "Bonjour" or brand-voice opener that the AI can't accidentally skip.

2. Quick Replies now support image and video attachments

Attach a JPG/PNG up to 5MB or an MP4/MOV up to 25MB to any quick reply. The media sends first, then your message and buttons — visual context lands before the customer even reads the text. Works on Facebook and Instagram DMs (images also work on WhatsApp). Great for product photos, demo clips, before/after shots, menus, and walkthrough screenshots.

3. Every conversation now gets auto-tagged by source

Comment Trigger, Visual Flow, Quick Reply, or CTWA Ad — every new conversation now carries the source tag automatically, so you can finally answer "which trigger captured this lead?" without digging through history. Tag chips and the conversation-list filter are now available on every tier (used to be Pro+ only).

Pro and above can override the system default with their own custom tag per trigger, flow, or quick reply — pick something like "Hot Lead" or "Black Friday" and it flows straight through to your Klaviyo or GHL segments.

4. New Manager team role — full dashboard access without sharing your login

Invite a teammate as a Manager (new role alongside Agent) and they'll log in with their own credentials and land on your full ChatGenius dashboard — Conversations, Captured Leads, Calendar, Content Studio, AI Config, Flows, WhatsApp Templates, Broadcasts, Comment Triggers, the lot. Billing, account settings, key rotation, and partner payouts stay owner-only. Promote or demote existing team members from Settings → Team. Available on Pro and above; Managers count against the same team limit as Agents.

Also shipped

• SMS Volume Pack — +6,000 outbound SMS per month, $79/mo add-on for Pro and Business clients with high-volume reminder, notification, or follow-up needs.
• Bundled checkout — new clients can now buy ChatGenius + Webhook Add-on + SMS Volume Pack together in a single Stripe Checkout, producing one subscription, one invoice, and one billing cycle instead of three.
• Team portal stress-tested for 30+ simultaneous agents — tightened heartbeat thresholds, capped inbox queries, added a composite index, and gated tab-hidden polling. Invisible at small team sizes; meaningful for any client running a real-team operation.

Plus team-portal polish: platform badges (Instagram/Facebook/WhatsApp/SMS/Telegram) are brighter and properly spaced, and the Online Team sidebar updates live when you change your status.

Four big upgrades this week. Plus a long polish list underneath.

1. AI Configuration got a full redesign

The AI Config page is now organized like Settings: a left sidebar with 8 focused sections grouped into Build, Extend, and Protect. Custom Rules & Guidelines is no longer hidden behind an "Advanced" accordion. Business Hours, Off-Limits Topics, and Solicitation Guard are all front-and-center now. Mobile got a complete rebuild too

2. Your AI just got sharper

Replies are now tighter (1-2 sentences), focused on one idea at a time, and won't fake actions like "I've passed this along." Markdown bullets, bold, and em dashes are stripped at the source. Onboarding scans also pull in more FAQs, services, and contact details out of the box.

3. No more dropped messages during high-volume bursts

When Meta's hourly cap gets hit during a viral post or comment-trigger storm, messages now queue and send when capacity opens up instead of getting dropped. Outbound limits are also raised closer to Meta's actual ceilings.

4. AI stops talking when a real person needs to step in

Once a conversation escalates, the AI stays silent across all 5 platforms until you take over. You'll also get an email if no teammate is available to pick it up, so escalations never go unnoticed.

Also shipped

• Explore before you connect - Build flows, triggers, quick replies, and content posts before connecting a channel.
• Auto-escalation toggle - Turn off AI-driven escalation in AI Config. Customer requests like "talk to a human" still work.
• Separate business contact - Set a customer-facing email and phone separate from your login/billing email.
• Daily Google Calendar sync - Edits made directly in Google Calendar now sync into ChatGenius automatically.
• Emojis save correctly - Emojis and special characters in AI Instructions no longer get replaced with question marks.
• Cleaner conversation list - No more flicker on poll, consistent badge sizing, and a distinct red badge for escalated chats.

Plus smaller fixes: escalation messages now match the conversation's language, the chat view stays pinned to the latest message while media loads, and the batch-window slider now has a 5-second minimum

Three big things shipped this week. Plus a lot of polish underneath.

1. TikTok publishing is live for everyone

TikTok publishing is now live for all Creator+ clients. Connect TikTok from the Connect page (same place you connect Instagram, Facebook, and WhatsApp), then schedule videos and photos to TikTok from Content Studio alongside your other platforms.

2. Redesigned Connect page (now with Integrations too)

The old one-long-scroll Connect page is now a clean two-column hub. Channel sidebar on the left, focused detail panel on the right. One click to reach any channel, live status dots, and a dedicated Plan & Billing and Help & Support section.

Google Calendar, GoHighLevel, and Klaviyo have also moved into Connect as a new Integrations group, so every service you connect to ChatGenius now lives in one place. The standalone Integrations nav entry is gone, and old links redirect automatically.

3. AI Media Library now supports videos

Upload MP4 or MOV files up to 25MB. When a customer asks for a walkthrough or demo, your AI sends the actual video instead of a placeholder. Works on Instagram and Facebook DMs.

Also shipped

• Quick Reply variations - One quick reply can hold up to 5 messages. ChatGenius rotates them randomly so repeat customers don't see the same canned reply.
• Instagram Live comment triggers - Auto-DM commenters during your live broadcasts using the same keyword triggers as regular posts.
• Unified Services - Your booking AI now pulls from the Service Menu in AI Config. Edit a service once, it's updated everywhere (price, duration, the whole thing).
• Simpler Knowledge Base limits - One unified cap covers FAQs, services, and documents instead of three overlapping ones.

Plus a pile of smaller fixes and polish: conversation actions no longer refresh the page, the Test Bot reliably produces real responses, WhatsApp claim actions behave correctly, and more.

Big week - WhatsApp just landed in ChatGenius.

WhatsApp is Here

Instagram DMs, Messenger, and now WhatsApp - all in one inbox. Same AI, same flows, same dashboard. Your customers can reach you on whatever platform they prefer, and you manage it all from one place.

What you get:

• Unified inbox - WhatsApp conversations live right next to your Instagram and Messenger threads
• AI-powered replies - The same AI that handles your DMs now works on WhatsApp too
• Template messaging - Create and send WhatsApp templates to re-engage customers after the conversation window closes
• Delivery tracking - See exactly whether your messages were delivered or failed, right on the message

Available on Professional and Business plans. Go to Settings to connect your WhatsApp Business Account.

Also This Week

• Faster AI responses - Optimized how the AI processes messages. Replies are noticeably quicker.
• PDF attachments in flows - Automatically deliver PDFs (pricing sheets, menus, guides) through your flow automations
• Comment trigger templates - 6 pre-built templates so you can set up common comment triggers in seconds
• Smarter service answers - AI gives shorter, more relevant answers instead of listing your entire catalog
• Real-time conversations - Messages now appear instantly with natural timing instead of all at once

Bug Fixes

• Fixed buttons becoming unclickable after switching portal sections
• Instagram AI replies now respect character limits instead of failing
• AI no longer awkwardly introduces itself in every reply
• Conversations stay scrolled to the latest messages when opened
• Flow media attachments (video, audio, PDF) now show in conversation history

Here's what's new with ChatGenius this week.

New Features

• Smoother Portal Navigation — Switching between sections is now instant with no page reloads
• Live Conversation Updates — New DMs appear in your inbox automatically within seconds, no refresh needed
• Test Bot Available on Free Tier — All users can now test their AI before connecting Instagram or Facebook
• OAuth Walkthrough Video — 60-second video showing exactly what happens when you connect your account
• Rebuilt Onboarding — Smarter website scanning, resumable progress, and a cleaner setup experience from start to finish

Bug Fixes

• Conversation filters (date, tags, source) no longer reset when the inbox auto-refreshes
• OAuth errors now redirect back to the connect page instead of the dashboard

Coming Soon

WhatsApp integration is on its way. Same AI, same dashboard — now in your customers' most-used messaging app. Stay tuned.

Big week. Here's what's new with ChatGenius.

New Features

• AI Comment Manager (Professional+) - AI now automatically replies to comments on your Instagram and Facebook posts using your knowledge base. Choose auto-reply or approval queue mode, set custom tone for public comments, and auto-hide spam. Find it in your sidebar under "Comment Manager."
• Community Hub - New section in your sidebar under "Explore." Browse and install flow templates shared by other businesses, publish your own flows to the community, and stay up to date with product updates, all in one place.
• Flow Export & Import - Download any flow as a .json file to back it up or share it. Import flows from files to quickly set up new automations. Look for the Export and Import buttons on your Flows page.
• AI Summary in Escalation Emails - When a conversation is escalated, the notification email now includes an AI-generated summary so your team can triage faster without opening the dashboard.
• Smarter Magic Wand - The AI Instructions magic wand now uses your full business context (name, tone, goals, knowledge base topics) to generate much better custom instructions. If you already have detailed instructions, it refines them instead of overwriting.
• Test Your Bot (Now Available) - The Test Bot widget in AI Config is now available to all Creator+ clients. Test your AI responses in real-time using your actual settings and knowledge base.

Improvements

• Flow Builder - Activating/pausing flows no longer reloads the page. Flows with delayed follow-ups (like Welcome Series) now work correctly. Help docs link added to the Flows header.
• Skeleton Loaders - Conversations, analytics, and Content Studio now show smooth loading animations instead of spinners.
• Reply Character Count - Live counter below the reply box adjusts per platform (IG 1000, FB 2000, SMS 160). Turns red and blocks send when over the limit.
• Quick Reply Preview - Preview button in the edit panel shows your message as it will appear to customers, with buttons rendered and placeholders replaced.
• Sync Status Banner - If your dashboard loses connection, a banner appears so you know your data might be stale. Auto-clears when reconnected.

Bug Fixes

• Fixed timezone always defaulting to Los Angeles for new signups. Your timezone is now detected correctly on signup and auto-corrects on next login.
• AI responses are now more accurate. Raised the similarity threshold to prevent irrelevant knowledge base content from being used.
• Follow-up sequences no longer end prematurely from false goodbye detection (e.g., "take care of my order" no longer triggers goodbye).
• Tag filtering now works correctly across all pages in conversation lists.
• Quick reply edit panel now correctly shows active/inactive status.

Coming Soon

TikTok content scheduling in Content Studio, more flow templates in the Community, expanded AI Comment Manager analytics, and additional automation triggers.

We just shipped something big.

Visual Flow Builder (Beta)

Build multi-step automated DM conversations with a visual drag-and-drop editor. Design the entire customer journey — from first message to booked appointment — without writing a single line of code.

What you can do:

• 12 node types — messages, collect input, conditions, delays, booking, AI steps, webhooks, and more
• AI-powered generation — describe your flow in plain English and AI builds the entire node graph
• Smart triggers — start flows from keywords, comment triggers, story replies, story mentions, shared reels/posts, or AI intent recognition
• Built-in analytics — see exactly where users drop off with per-node heatmaps
• Test before going live — walk through your entire flow in a chat simulator without sending real DMs

How to get started:

1. Open your ChatGenius dashboard and click Flows in the sidebar
2. Click Create Flow — pick a template or start from scratch
3. Drag nodes onto the canvas, connect them, and hit Test to preview
4. When you're ready, activate the flow and it starts working immediately

New Features

• Partner/Affiliate Program — Refer new clients with your unique link and earn 30% recurring commission on every payment. Track referrals, commissions, and payouts from your partner dashboard
• Instagram Post Share Trigger — New event trigger that fires when someone shares an Instagram post in your DMs. Set up an automated response for shared posts just like you can for reels

Bug Fixes

• Better image handling — When a customer sends an image and your plan doesn't include image processing, the AI now replies with a helpful message asking them to describe what they're showing instead of a confusing error
• AI respects boundaries — When a customer says "no" to booking or a service, the AI now accepts immediately and moves on instead of re-offering repeatedly
• Faster AI responses — Eliminated unnecessary retry logic that was doubling response times
• Smarter AI tone — Removed filler phrases, more natural fallback when AI doesn't know something
• Team invitations fixed — Invite, activate, and manage team members without issues

Here's what's new with ChatGenius this week.

New Features

• CSAT Satisfaction Surveys — After a conversation is resolved, ChatGenius can now automatically send a 1-5 star satisfaction survey right in the DM. Configure it in AI Config under the CSAT section.
• Solicitation Guard — The AI now auto-detects sales pitches and spam DMs, sends a polite dismissal, pauses the conversation, and flags it so your team isn't wasting time on solicitors.
• Assign Conversations — You can now assign conversations to specific team members directly from the Conversations tab. The agent picker shows online/offline status.
• AI Response Feedback — Rate any AI response with a thumbs up or down directly in the conversation view. Your feedback helps us improve AI quality over time.
• Image Attachments in Comment Trigger DMs — You can now attach an image to any comment trigger DM, sending a rich card with your image, message, and buttons in one message.
• Auto-Reply to Collab Invites & Unrecognized Media — New event trigger in Quick Replies lets you auto-respond when someone sends a collaboration invite or other unrecognized message type. Set up from Quick Replies → New Quick Reply → Trigger Mode: Event.

Improvements

• Streamlined Connection Flow — Connecting Facebook and Instagram is now faster. Single-page accounts auto-connect instantly after granting permissions.
• Knowledge Base Storage Limits — Limits are now based on total storage size instead of document count, giving you more flexibility. Check your usage in AI Config.
• Comment trigger DMs no longer show the opt-out footer, making automated replies feel more natural and personal.

Bug Fixes

• Fixed a timezone issue where portal timestamps could display 1 hour off during daylight saving time
• Fixed PDF attachments sent in DMs not being processed by the AI

Here's what we've shipped recently across ChatGenius.

New Features

• AI Media Library — Upload images your AI can contextually send in conversations (Creator+ plans)
• Conversation Tags — Organize and label conversations for faster follow-up (Pro+ plans)
• Direct Agent Transfer — Hand off conversations to specific team members + new team performance dashboard
• AI Blocklist — Block specific users from receiving AI responses
• Two-Way Support Tickets — Reply to support tickets directly with threaded conversations
• Link Click Tracking — See when customers click URL buttons in comment trigger DMs

AI Improvements

• Fixed AI hallucinating booking URLs (e.g. fake Calendly links) — it now only shares real links you've configured
• Fixed AI hallucinating actions outside its actual capabilities
• Prevented follow-up messages from contradicting the original AI response
• RAG knowledge base improvements — better chunking and multi-document support

Bug Fixes & Polish

• Quick Replies section completely redesigned
• Comment Triggers section completely redesigned
• Mobile responsiveness improvements across Settings, Team Management, Calendar, AI Config, and Content Studio
• Fixed overview conversations not opening the correct thread
• Fixed View Lead button linking to the wrong conversation
• Comment Triggers: fixed post picker issues for accounts with emoji in captions

We're shipping updates every week. If something isn't working right or you have a feature request, just reply to this email.