TLDR: When your chatbot hands a chat to a human, the teammates you pick now get a browser alert. The Inbox has a cleaner look that works on phones. Contains keywords now match whole words only, so fewer wrong replies.

Here's what's new 👇

🙋 Takeover Alerts for Your Team

The problem: When your chatbot passed a chat to a human, only people linked on Telegram were told. Your team in ReplyLa could miss it.

• Go to your chatbot's Human Takeover, open Notify, and tick the people you want under Team members. ✅
• Each person shows a status: Will be alerted, Not set up or Alerts off, so you know who will really get it.
• What they get shows a sample of the alert before you save.
• Telegram still works. It now sits under Also send to Telegram.
• The alert says which customer needs help. Click Open chat to jump straight in.
• A picked teammate who has not allowed browser alerts sees You are assigned for takeover until they allow them.

🔔 Simpler Alert Settings

The problem: The alert settings had many lines to tick, and it was not clear what each person would get.

• Settings, Notifications now has two choices: Only assigned to me or All messages.
• With Only assigned to me, you can also tick Chatbot is off and New customer.
• A WhatsApp channel disconnects has its own place under Channels.
• Takeovers you are picked for, and replies in chats assigned to you, always alert you.
• The owner starts on All messages. Everyone else starts on Only assigned to me.

💬 A Roomier Inbox, Even on Your Phone

The problem: On a phone, a warning banner could push the typing box off the screen. The chat list was crowded too.

• The typing box always stays on screen, even when a banner shows.
• The typing bar is now one neat card. Tap the add button for File, Photos & Videos, File Manager, Location or Template.
• The chat list shows the customer's name only, in tidy rows of the same height. A small dot in the corner shows if the chat window is open.
• Labels show newest first. In the chat header, extra labels fold into a +N you can tap to see them all.
• On phones and tablets, a bottom bar gives you Home, Filter, Search and your devices. Labels and Assign live under More options.

🔑 Keywords Match Whole Words

The problem: A Contains keyword could fire inside other words. For example, the keyword "A" matched "salam", so the wrong reply was sent and it looked like spam.

• Contains keywords now match only whole words or phrases. ✅
• Other forms of a word no longer match. "harga" will not match "harganya", and "minat" will not match "berminat".
• If you want those to match too, add them as extra keywords.
• Exact keywords work the same as before.

🛠 Bug Fixes

• Template links sent once. Some keyword replies with a template sent the start of a link twice when the customer's details held the full link. Now the link is sent once and correctly.
• Closing rate chart looks right. The closing rate chart dropped to 0% on days with no data. Now it skips those days.
• Better security. We tightened security so your data stays private to your own organization.

TLDR: Browser alerts now show who messaged you and what they said, and you choose which alerts you get. You can pick contacts who came from an ad or still have an open chat window. Template links now ask only for the ending.

Here's what's new 👇

🔔 Alerts That Tell You Who Wrote

The problem: Browser alerts only said "New messages" for a device, so you had to open ReplyLa to see who it was. And you could not choose which messages should alert you.

• Each customer message now gets its own alert with the customer's name and the start of their message. ✅
• Click the alert, or Open chat, to jump straight to that chat.
• Photos, videos and files show as "Sent Photo" and so on. Captions never appear on your lock screen.
• Go to Settings, Notifications, Alerts for your business, and tick what you want under Alert me about: All messages, Messages in chats with the chatbot off, A new customer starts a chat, A customer replies in a chat assigned to me or A WhatsApp channel disconnects.
• Prefer privacy? Switch off Show message text in alerts and you only see who wrote. 🔒
• Already had alerts on? Every line is ticked for you and message text is off, until you change it.
• On a shared computer, the next person who logs in does not see the last person's customer alerts.

🎯 Three New Ways to Pick Your Audience

The problem: You could not easily find the customers who came from your ads, or the ones you can still message without a template.

• Contacts and Blaster now have a Reach group of filters: Came from an ad, Reply window open and Free window open.
• Window filters are checked on the device you send from. On the last step, Blaster recounts your audience for that device so the number is right.
• If nobody on that device matches, Blaster stops you from sending to an empty list. You can still save it as a draft.
• Scheduling a blast with a window filter? Blaster reminds you that windows are checked when you save.

🔗 Template Links: Type Only the Ending

The problem: When a template message had a link with a changing ending (like a tracking number), you had to type the whole link, and it was easy to get it wrong.

• The fixed part of the link now sits in front of the box with a Link ending badge. You only type the part that changes.
• Paste a full link and we keep only the ending for you.
• A Sends: line shows the final link before you save.
• This works in Blaster, keyword replies and the template builder.

🛠 Bug Fixes

• Blasts no longer go to everyone by mistake. If every contact you ticked had been deleted before you saved, the blast picked everyone who matched your filters. Now it stops and tells you those contacts are no longer available.
• Saved drafts show the right audience. Opening a saved blast draft showed 0 recipients. Now it shows the real number.
• Very big contact picks get a clear message. Ticking a huge number of contacts used to fail with an error. Now you get a clear message instead.
• Links sent correctly. Some template messages sent the start of a link twice, and Meta rejected some templates for the same reason. Now the link is sent once and correctly.
• No more double slashes. Typing a link ending by hand could add an extra slash to the link. Now it saves exactly what you mean.

TLDR: Your closing rate now counts only the leads you marked as Closed, the same way everywhere. Analytics now shows an estimate of what WhatsApp charges for your blasts. Your chatbot sends pictures only when you switch it on, and you can find a chat by the customer's WhatsApp username.

Here's what's new 👇

📊 A Closing Rate You Can Trust

The problem: Your closing rate used to guess. If you had not chosen your Closed labels, it counted whatever was in the last step of your funnel, which was often the wrong thing.

• A lead now counts as closed only when it has one of your Closed labels from Auto Label.
• Your Dashboard, the Analytics page and your AI assistant now all count closed leads the same way.
• No Closed labels yet? The card shows Set your Closed labels with a button that takes you straight to Edit Funnel.
• Your closing rate may change after this update. That is normal: it now shows the real number.

💰 See What Your Blasts Cost

The problem: You could not see how much WhatsApp charges for the messages you send.

• The Blasts tab in Analytics now shows Cost by pricing type and Per-campaign cost & leads.
• These numbers are an estimate, based on how many messages Meta delivered and Malaysia's WhatsApp prices.
• This works for WhatsApp devices connected through Meta.

🖼 Pictures Follow One Simple Switch

The problem: It was not clear when your chatbot would send a picture or video, and when it would not.

• Send Images & Videos in AI Settings now decides it on its own. On: your chatbot can send your files. Off: it replies in text only.
• If you already added files to your chatbot, we switched this on for you, so nothing stops working.
• If the switch is off but your chatbot has files, AI Settings now warns you.
• Your keyword replies and follow up messages still send their pictures as before.
• AI Checker now runs only if you switched it on. Chatbots without it no longer spend tokens on it.
• Your chatbot now uses fewer tokens for the same chats. 💸

🔍 Find a Chat by WhatsApp Username

The problem: Some customers show a name instead of a phone number, and you could not search for them by their WhatsApp @username.

• Type part of a customer's @username in the Inbox search to find their chat.
• Your AI assistant (Claude or ChatGPT) can find them the same way, and can now tell which part of your chatbot sent each message, like a keyword rule or a follow up.

🛠 Bug Fixes

• One chatbot, one device. Connecting a chatbot from the device page sometimes left it on its old device too. Now it moves cleanly to the new device. If you use webhooks, you also get a new event when a chatbot leaves a device.
• STOP works for every customer. When a customer with no visible phone number replied STOP, the chatbot answered instead of unsubscribing them. Now they are unsubscribed properly.
• You pay the exact price shown. Some checkouts charged one sen less than the price on the page. Now the amount always matches.
• Guest checkout creates your account. Buying without an account and ticking Remember my card sometimes took payment but did not create the account. Now it does, and stuck payments are being fixed.
• Free coupons at guest checkout work. A 100% discount coupon now finishes the order properly, and a one-time coupon cannot be used twice.
• Account Health tells the truth. The Advanced tab sometimes said "Blocked" while your device was still replying. Now it matches the status at the top of the page.
• Numbered template variables get examples. Templates using {{1}}, {{2}} now show example fields, so Meta can review them.
• Campaign export follows your filters. Exporting campaign recipients now uses the search and filters you have on screen.
• Better security. We tightened security so your data stays private to your own organization.

TLDR: Website links now start with https:// for you, so you only type the rest. Blasts with a website button that changes per customer now send properly. When your chatbot passes a chat to your team, it now really stops and tells your team.

Here's what's new 👇

🔗 Easier Website Links & Buttons

The problem: Website links were easy to type wrong. A missing or double https:// could break your button.

• https:// is now already there. You only type the rest of the link.
• This works for template website buttons, your device's business website, and page links in AI Knowledge.
• If you paste a full link, we save it once. No double https://.
• Website buttons now have two choices: Fixed (same link for everyone) or Dynamic (the end of the link changes for each customer, like an order number).
• Your old templates still work. You do not need to change anything.
• On a phone, the template page buttons now fit nicely on screen.

📊 Campaign Report

The problem: When some blast messages failed, the reason was hard to find.

• Error breakdown is now near the top, right under Sending progress.

🛠 Bug Fixes

• Blasts with Dynamic buttons work again. These blasts were rejected by Meta because the end of the link was empty. Now the blast setup asks you to fill it in before you can save.
• Team handover works. Sometimes the chatbot knew a human should take over, but kept replying and did not tell your team. Now it stops and tells your team. You may get more handover alerts than before.
• No more double messages. Sometimes a customer got the same message twice. This will not happen now.
• Unsupported messages now show. Some message types used to show as an empty space in the chat. Now you see "Unsupported message", and the chatbot does not reply to it.
• No more wrong phone numbers. Some devices showed a phone number that was not real. Now it shows empty.
• Better security. We made security stronger so your data stays private to your business.

TLDR: The Token Usage page now shows your balance in one clear card, and the "% used" warning counts from your last top-up, so it stops scaring you right after you buy more. ReplyLa now asks you to turn on browser alerts, lets you name each browser, and works on iPhone from the Home Screen. Your connected AI assistant can also help you answer the questions your chatbot was unsure about.

Here's what's new 👇

🪙 A Clearer Token Usage Page

The problem: The Token Usage page was busy and hard to read at a glance. The "% used" figure counted every token you ever had, so right after a top-up you could still see "Tokens almost depleted" and get warning emails.

• The top of the page is now one balance card: tokens left in big numbers, one bar split by where your tokens came from, and a simple breakdown underneath. ✅
• "% used" now counts from your last top-up, plan renewal or plan change, so a fresh top-up starts you near 0% used. 📉
• The Subscription page shows "X of Y left since your top-up" for your AI Tokens.
• If you closed the 75% or 90% warning before, it comes back for your next fill-up, so you never miss it when it matters.
• Usage history opens on Est. cost in ringgit. Tap Show token breakdown if you want the token numbers too. 💰
• Token History now lists your top-ups and refills only, so it is easier to follow.
• Big numbers and long contact names fit neatly instead of spilling out.

🔔 Browser Alerts Are Easier to Turn On

The problem: Browser alerts for new messages were hidden in Settings, so many of you never knew they existed. You could not tell your browsers apart in the list, and iPhone was not supported.

• A small Turn on alerts card now appears at the bottom right of your dashboard. One tap and you are set. ✅
• Tap Not now and it waits a week before asking again. After the third time it stops asking.
• Give each browser a name, like "Office laptop" or "My phone", in Settings, Notifications. ✏️
• iPhone and iPad are now supported. On iOS 16.4 or newer, add ReplyLa to your Home Screen from Safari, open it from there, and turn on alerts. 📱
• Team members who cannot see the Inbox are not asked to turn on alerts.

🤖 Your AI Assistant Can Help With Unsure Questions (Early Access)

The problem: Questions your chatbot was not sure how to answer could only be reviewed by hand on the AI Feedback page.

• A connected Claude or ChatGPT can now show you the questions your chatbot was unsure about, newest first. ✅
• It can suggest an answer and save it to your AI Knowledge as a draft. Nothing goes live until you publish.
• The AI Feedback page shows an Answer in draft badge on those questions, so you and the assistant never answer the same one twice. 🏷️
• The setup guide your assistant uses now covers more types of business, including services sold through a consultation and services done at the customer's place.

🛠 Bug Fixes

• Apostrophes stay as apostrophes. Answers saved from the AI Feedback page could reach your chatbot with a strange code in place of an apostrophe, so "don't" looked garbled. They now read exactly as you typed them.
• Inbox photos and files load more reliably. When Meta was slow or unreachable, photos and files in the Inbox could hang or say "not found". They now give up quickly and cleanly instead of slowing everything down.
• Readable brand buttons. On the Yellow colour theme, some buttons in CRM, Add Contact and Checkout had text that was hard to read. The text now stands out clearly. 🎨

TLDR: Chats that start from a Click-to-WhatsApp ad now show the ad itself right in the Inbox, so you know exactly what your customer saw. Connecting Claude or ChatGPT is quicker with simple access presets. Team members can open Contacts, Chatbot and Device Manager again.

Here's what's new 👇

🎯 See the Ad Behind Every Lead

The problem: When a customer messaged you from a Facebook or Instagram ad, the Inbox did not show which ad they tapped. You had to guess what they saw before they said hello.

• The first message from an ad lead now shows a From ad card with the ad's picture, headline, text and platform. ✅
• Tap View ad to open the exact ad in a new tab. Chats that start from a post show From post instead. 📣
• The contact panel on the right shows an Ad ID row with a View ad link for every contact who came from an ad.
• Ad pictures are kept by Meta for about five days. After that the picture is hidden, but the card and its details stay.
• This works on WhatsApp channels connected through Meta. Unofficial channels do not receive ad details, so no card shows there.

🔌 A Simpler AI Connect Screen (Early Access)

The problem: Connecting Claude or ChatGPT to your business showed a long list of permissions to tick one by one, with little help on what each one meant.

• Pick one of four presets: View only (recommended), View and edit, Full access or Custom. ✅
• A live summary beside the presets shows what your assistant can do in each part of ReplyLa.
• Every permission now has a plain explanation, and risky ones are tagged, like "Messages customers", "Uses AI tokens" or "Goes live". ⚠️
• View and edit never sends messages, deletes anything, trains, or publishes your chatbot.
• On a phone the screen stacks neatly, with Cancel and Connect always at the bottom. 📱

🛠 Bug Fixes

• Team members can open their pages again. Some team members saw an error when opening Contacts, Chatbot or Device Manager. These pages now load normally for every role.
• Saving roles is lighter. Saving custom roles and their permissions now works smoothly without slowing the system down.

TLDR: A connected Claude or ChatGPT can now publish your chatbot changes and train its AI Knowledge, but only after you say yes. A new Publish history shows who published each version of your chatbot. Chatbot tags now show as labels in your business, and typing a tag in different capitals no longer makes a duplicate.

Here's what's new 👇

🤖 Your AI Assistant Can Finish the Job (Early Access)

The problem: Your connected assistant could prepare chatbot changes as drafts, but you still had to open the dashboard to publish them, train the chatbot or add a PDF. Many chatbot pages were also out of its reach.

• It can now show you a summary of what changed, then publish the draft once you say yes. It can also throw a draft away if you change your mind. ✅
• After you add new knowledge, it can train your chatbot on it. It tells you first that training uses AI tokens, and only starts when you agree. 📚
• It can add a PDF from your File Manager to your chatbot's AI Knowledge. Your original file in File Manager is never touched.
• It can now read and adjust AI Tools, Keywords, Auto Label, Auto Follow Up and Human Takeover, plus your chatbot's name and tags. Every change still waits as a draft until you publish.
• It can switch your chatbot between Guided and Custom mode, after showing you the same warning the dashboard shows.
• It can tell you which chatbot features your plan does not include, so you know what an upgrade unlocks.
• It can explain the most common chatbot problems and point to the setting that fixes each one. 🔧
• Publishing and training are switched off for every assistant until you allow them. Open your assistant's Edit Access and tick the new publish and train permissions to turn them on. 🔌

📜 Publish History

The problem: There was no way to see who published each version of your chatbot, or whether a change came from a teammate or an AI assistant.

• Open Publish history from the AI Playground header to see your latest 20 published versions. ✅
• Each one shows who published it, a person on your team or a named AI assistant.
• Versions published before this update say "Published before publishers were recorded".

🏷️ Chatbot Tags Are Now Labels

The problem: Chatbot tags lived apart from your labels, so the same name could exist twice and a tag could change on its own when a label was renamed.

• Your chatbot tags now belong to your own business's labels. You may see a few new labels on the Labels page with 0 contacts. These are your chatbots' existing tags, nothing was added by mistake. 👀
• Typing a tag in different capitals, like "vip" instead of "VIP", now joins your existing label instead of making a second one.
• In CRM, the AI icon now only shows on labels your Auto Label rules can actually add.

🛠 Bug Fixes

• Auto Label saves again. A label name with an accented capital letter, like ÉTÉ, made every save of the Auto Label page fail. It now saves normally.
• No more re-publishing an old version. Clicking Publish on a page left open for a while could publish a version that was already live. It now stops and asks you to refresh.
• Publish and discard at the same time. Discarding a draft at the exact moment someone published it showed an error. Both now finish cleanly.
• AI Tools drafts save properly. In Playground mode, saving or switching some AI Tools said "Draft saved" when nothing was saved. Your change is now kept in the draft.
• Fewer false "live settings changed" warnings. Some drafts were blocked from publishing even though nobody had changed the live chatbot. That no longer happens.
• Tighter security. We tightened security so your data stays private to your own organization. 🔒

TLDR: Opening a chat in the Inbox now marks it read for your whole team and gives your customer blue ticks, plus Mark as unread and a private mode. Owners can pause the chatbot for one customer straight from WhatsApp. The reply timer stops calling the 24-hour window free.

Here's what's new 👇

👀 Read and Unread in the Inbox

The problem: Opening a chat on the web never marked it as read, so the same chats stayed unread for everyone. Your customer also never saw blue ticks, even after your team had read their message.

• Opening a chat now marks it read for your whole team. A new message that arrives while the chat is on your screen is marked read too. ✅
• On WhatsApp channels connected through Meta, your customer sees blue ticks, so they know their message was seen. 💙
• Use Mark as unread in the chat menu to put a chat back in Unread for the team. It shows when the latest message came from the customer.
• Turn on Private mode to open chats without marking them read. It is your own setting and follows you across every business you belong to. 🕶️
• Unread chats in the list now show a dot and a bolder name.

📱 Pause One Customer From WhatsApp

The problem: Owner commands could only stop the whole chatbot. There was no way to silence it for just one customer without opening the dashboard. Owners whose number was saved with a "+" in front were ignored completely, so the AI answered their commands like a normal message.

• Send /stop followed by a customer's number, for example /stop 60123456789, to pause the chatbot for just that customer. Send /start with the number to turn it back on. ✅
• The pause covers all of that customer's chats with your business, on every channel, the same as switching the AI off in the Inbox.
• Turning the chatbot back on also resets that customer's reply limit.
• You get a clear reply saying what happened, or what to fix if the number was wrong.
• Owner numbers cannot be paused, so you can never lock yourself out. 🔒
• Owner numbers saved with a "+" in front now work.

⏱️ An Honest Reply Window Timer

The problem: From 1 October, WhatsApp charges for replies inside the 24-hour window, but the Inbox timer still called that window free.

• The timer and its "How replies work" popup no longer call the 24-hour reply window free. ✅
• For a chat from a Click-to-WhatsApp ad, the timer shows exactly what is free and until when, using WhatsApp's own end date. That free period can now last up to 7 days. 📅
• Before WhatsApp confirms the free period, the timer shows how long you have left to reply to make it free.
• The "Free message closed" error is now called "Reply window closed".
• When WhatsApp has not confirmed how long an ad lead's free period lasts, we now count it as 24 hours instead of 72, so you are not charged by surprise. Ad Leads Only follows the same rule.

✂️ Split Replies for Ad Customers

The problem: Turning off Auto-Split Messages also stopped split replies for ad leads, even while their messages were free.

• Under Auto-Split Messages in AI Settings there is a new Auto-Split for Ad Customers switch. ✅
• With it on, ad leads still get normal multi-message replies while their free period is open, even when Auto-Split is off for everyone else.
• Once the free period ends, replies go back to one message. If we are not sure the period is still open, we play safe and send one message.

🤖 Your AI Assistant Can Do More (Early Access)

The problem: A connected Claude or ChatGPT could read chats and reply, but it could not see the bigger picture of your Inbox, your campaigns or your chatbot setup.

• It can give you an Inbox summary: unread chats, chats waiting for a reply, chats that need a human and reply windows about to close. It can also mark chats read, assign them to a teammate and switch the chatbot on or off for a chat.
• It can read your Analytics, like your closing rate and where your leads come from. 📊
• It can list your leads by the ad they came from and by date. 🎯
• It can view your Blaster campaigns and create new ones as drafts. It never sends a campaign. 📢
• It can read and change your chatbot settings and AI Workflows, and explain why a chatbot is not replying. Every change waits as a draft for you to review and publish.
• It can see the images in your File Manager and add them to your chatbot.
• Already connected? Open your assistant's Edit Access and tick the new permissions to switch these on. 🔌

🛠 Bug Fixes

• AI Workflows shows your device. The AI Workflows page said no device was connected even when one was. It now shows the right device.
• Loading spinner on confirm boxes. When you confirmed an action like a delete, the loading spinner never appeared. It now shows while the action runs.
• Tighter security. We tightened security so your data stays private to your own organization. 🔒

TLDR: Turn on Auto Reload and your AI tokens top themselves up from your default card before they run out, with limits you control. A new Ad Leads Only setting lets your chatbot answer only people who came from your Click-to-WhatsApp ads and hands everyone else to your team. The Blaster also stops getting stuck when a campaign cannot go out.

Here's what's new 👇

🔋 AI Tokens That Top Themselves Up

The problem: When your AI token balance ran out, your chatbot stopped replying until someone noticed and bought more tokens by hand. Customers could go unanswered for hours.

• In Subscription, Token Usage, there is a new Auto Reload card. Switch it on and choose the balance that triggers a top up, how many tokens to add each time, a monthly spending cap in RM and how many top ups are allowed per day. ✅
• The cost of each top up is shown before you save.
• Top ups charge your default card, the same card your renewals use. You get the same receipt and invoice as a normal purchase, marked "Automatic". 🧾
• You are never charged twice for one top up. If a payment result is unclear, we check with the payment provider before adding tokens.
• You get a notification when a top up succeeds, when one of your limits is holding it back, or when your card is declined. A declined card is not tried again until you fix it.
• It stays off until you turn it on. If you have no saved card we can charge, the card tells you how to add one.
• Changing your default card now asks you first, because the same card also pays for automatic top ups.

🎯 Ad Leads Only

The problem: From 1 October, WhatsApp starts charging for more of the customer service messages a business sends. Leads who click your Click-to-WhatsApp ads get a free window, and many businesses only want the AI answering then. There was no way to set that.

• In your chatbot's General Settings, turn on Ad Leads Only in the new "Handover to Your Team" card. ✅
• Leads from a recent Click-to-WhatsApp ad keep getting normal AI replies.
• Anyone else gets one message you write, then the AI pauses for that contact and your team takes over.
• Your team gets a bell notification ("Lead handed to your team") and a note on the conversation. 🔔
• The contact appears in your human takeover list. When your team resumes the AI, it stays on for that contact.
• It is off by default and works on WhatsApp channels connected through Meta.
• Conversation Limit now sits in the same "Handover to Your Team" card.

🛠 Bug Fixes

• Campaigns over your plan's limit stay editable. Scheduling a Blaster campaign with more recipients than your plan allows showed an error page, even though the campaign was saved. It now stays as a draft and tells you why.
• No more campaigns stuck on processing. Creating a campaign with an audience that had failed to build left the campaign stuck forever. You now see a clear error instead.
• Returning ad leads show the right window. When a past customer clicked your ad again, the Inbox window chip could say the free window was closed. It now shows it as open.
• A clearer error label in campaign reports. One WhatsApp error was labelled wrongly in the campaign error summary. It now describes the real reason.

TLDR: Choosing a default card now really changes which card pays for your renewals, and a new Payment Methods page puts all your cards in one place. The Inbox chat looks cleaner, with grouped bubbles and a colour you can pick for each channel. You can also turn off auto-split replies, and AI follow ups stop using tokens once you run out.

Here's what's new 👇

💳 Your Default Card Is Now the One We Charge

The problem: Pressing Set Default on a card looked like it worked, but nothing was saved. A renewal could charge a different card from the one you picked, a card you removed could come back, and some valid cards showed as expired.

• Set Default now sticks, and your plan and add-on renewals charge exactly that card. ✅
• We never charge a card you did not choose. If you have no usable default card, the renewal stops and you are told, instead of us picking one for you.
• A new Payment Methods page in the sidebar lets the account owner add cards, pick the default and remove cards.
• Removing your default card no longer quietly makes another card the default. If no usable card is left, auto-renewal switches off and you get a notification, so you can add a new card.
• A card you removed stays removed, and cards your payment provider has deleted drop off your list by themselves.
• A card that was declined once for low balance is no longer blocked for good. It can be charged again on the next try.
• Valid cards no longer show as expired.
• After adding a card you go back to the page you started from. If you cancel on the card form, you see "No card was added" instead of a payment error.
• Only the account owner can add cards or change the default card. 🔒

💬 A Cleaner Inbox Chat

The problem: Every message in a busy chat carried its own time and status, so long conversations were hard to scan. Every channel also looked exactly the same.

• Messages sent back to back now group together, with one line under the last bubble showing the status, the time and who sent it.
• Pick a bubble colour for each channel with Colour theme in the chat header, so you can tell your numbers apart at a glance. 🎨
• A failed message shows a red icon with the reason. Click it to retry, and we ask first so you never resend by accident.
• A reply that was stopped because the chat was paused shows a grey pause icon, so you know it was not sent.
• Pinch on your trackpad or phone to make chat text bigger or smaller.
• The messaging window timer now opens a short explainer of the 24-hour rule and Click-to-WhatsApp ad windows, right inside ReplyLa. It only shows on WhatsApp channels where it applies.

✂️ Turn Off Auto-Split Replies

The problem: The AI always broke its replies into several short messages, and there was no way to switch that off.

• In AI Settings, Extra Features, there is a new Auto-Split Messages switch. It is on by default, so nothing changes unless you turn it off. ✅
• With it off, the AI sends one message instead of several. A photo or file still goes as its own message.
• The setting now also applies to follow ups and to replies that include an image.
• Very long AI replies and captions used to be rejected by WhatsApp and never arrive. They are now kept within WhatsApp's limits, so they always get delivered. 📬

🪙 AI Follow Ups Stop When You Run Out of Tokens

The problem: When your AI tokens ran out, normal replies stopped, but scheduled AI follow ups kept going and kept using tokens, pushing your balance further below zero.

• Follow ups written by the AI are now cancelled once you are out of tokens. ✅
• Follow ups you wrote yourself and template follow ups still go out as normal.
• A cancelled follow up shows an "Out of AI tokens" label in Activities, so you know why it did not send.
• Chatbots that use your own AI key are not affected.

🛠 Bug Fixes

• Tighter security. We tightened security so your data stays private to your own organization.