Releases

Bulk Assign, the tool you already use to set status, agents and departments on many conversations at once, now also marks conversations read or unread.

📥 Two new bulk actions

• Mark as Read, for clearing a backlog of conversations that no longer need attention.
• Mark as Unread, for flagging a batch your team still has to work through.

Both run on the same selection you already make for status and agent changes, and they work on large selections rather than a page at a time, so a queue of dozens goes in a single pass.

⚡ Your inbox updates as it runs

Conversations change state in the open inbox while the bulk action runs, on desktop and in the mobile app, so you are not left refreshing to find out whether it worked. Read receipts on the conversations you mark are unaffected and behave exactly as they do when you open a conversation yourself.

✅ How to use it

1. Open Bulk Assign.
2. Filter to the conversations you want and select them.
3. Choose Mark as Read or Mark as Unread and apply.

You can now pull a teammate into a conversation by name. Type @ in a private note and pick the person you need; they get a notification that takes them straight to the thread.

💬 Where you can mention someone

• Inbox private notes, the main one. The note stays internal; the customer never sees it.
• Contact notes in the Contact Hub, including when you edit an existing note.
• Tickets, in the ticket form and description, whether the ticket came from a single message or a whole conversation.

👥 The picker only offers people who can help

Start typing after the @ and search by first name, last name, or both. Each suggestion shows the teammate's email and role next to their name, which settles it when two people share a first name.

The list is limited to teammates who can actually open that conversation, so you cannot mention someone into a thread they are not allowed to see, and every internal teammate on your workspace is reachable.

🔔 The notification

The person you mention gets an in-app notification reading who mentioned them and where. Mentions are never bundled together with other alerts, so five mentions are five notifications rather than one summary you can miss.

✅ How to use it

1. Open a conversation and switch to the private note box.
2. Type @ followed by part of the person's name.
3. Pick them from the list and finish your note.

Mentions also come through on the mobile app and in the API, rendered as plain readable names.

Meta's own AI assistant can now answer customers on your WhatsApp number, and you set the whole thing up inside ChatMaxima. The agent replies first, your inbox stays in sync, and your team can take over any conversation at any point.

🤖 Set it up on a number

Go to Settings, WhatsApp, and open Actions on the number you want. Meta Business Agent shows whether the number is eligible, walks you through creating the agent, and shows an On or Off badge from then on.

• Who it replies to: everyone, or only a list of test numbers you add, which is the safe way to trial it.
• Handoff to your team, so the agent transfers the conversation when a customer asks for a person or it cannot help. Use Meta's wording, let the agent phrase it, or write your own.
• Follow-ups for customers who go quiet, from 5 minutes to 24 hours.
• Never say, a list of phrases the agent will avoid.

📚 Teach it your business

The Knowledge tab is where the agent learns what you sell and how you work.

• FAQs, the questions customers actually ask, with your answers.
• Websites, which Meta crawls in the background and keeps as reference. Crawl status and page counts show on the tab.
• Files: PDF, DOC, DOCX, PNG, JPG, CSV or XLSX, for price lists, policies and specifications.

The Skills tab goes further, letting you write named instructions that shape how the agent behaves in a specific situation: a returns skill, a booking skill, a tone-of-voice skill. Each one says when it applies and what to do.

🛠️ Let it do things, not just answer

The Actions tab turns on what the agent can do mid-conversation, using integrations you have already connected:

• Contacts and tickets: look up a contact, save or update one, open a support ticket.
• Shopify and WooCommerce: order status with tracking, and product search with price and availability.
• Google Calendar: check free slots and book an appointment with the customer invited.
• HubSpot and Zoho CRM: find a contact, or create a new lead.

Your integration credentials never leave ChatMaxima. Meta's agent only ever holds a ChatMaxima key and asks us to run the action on its behalf.

💬 Try it before a customer does

The Test chat tab runs the full agent pipeline in the modal. It reaches no real customer and is not billed as a conversation, so you can rehearse the awkward questions, see when it hands off, and try the quick replies as buttons.

✅ How to use it

1. Accept the Meta Business Agent terms for the number in WhatsApp Manager.
2. In ChatMaxima open Settings, WhatsApp, Actions, Meta Business Agent.
3. Set up the agent, add your FAQs and websites, and switch on the actions you want.
4. Add your own number as a test number, set the audience to Only test numbers, and switch the agent on.
5. When it reads well, change the audience to Everyone.

Availability is decided by Meta and depends on your business country and vertical, so the panel tells you plainly if your number is not eligible yet.

Available in all 11 dashboard languages.

Inbox translation used to run into English, whatever language you work in. You can now choose the language your customer messages are translated into, and it sticks across every conversation you open.

🌍 Pick your own translation language

The picker sits in the profile menu at the top of the dashboard, next to the language picker for the dashboard itself. They are deliberately separate: one changes the buttons and menus you see, the other changes what incoming messages get translated into.

• 71 languages to translate into, far more than the 11 the dashboard interface ships in.
• Set once, applies everywhere: auto-translate, the per-message Translate action, and translation inside a ticket all follow the same choice.
• Sensible default: if you have not chosen one, translation follows your dashboard language, so an agent working in Arabic gets Arabic with no setup at all.

Changing it does not reload the page, because nothing on screen is written in it. It only changes what the next translation aims at.

✅ How to use it

1. Open the profile menu at the top right of the dashboard.
2. Choose your language under Inbox translation. The list is searchable.
3. Open any conversation and translate a message.

The Make this my default option inside the auto-translate window still works and writes the same setting, so the two stay in sync.

🔌 Existing integrations are unaffected

Calls made with an API key keep translating into English unless they pass a target language explicitly, so nothing you have already built changes behaviour.

ChatMaxima now speaks MQTT. Connect your broker once under Integrations, then drop an MQTT Publish block into any flow to send a message to a topic: a device command, a reminder, a queue entry for another system to pick up.

📡 Connect your broker

Add an MQTT integration with your broker host, port, and an access credential that has publish permission. HiveMQ Cloud, Mosquitto and EMQX are all supported.

• TLS by default on port 8883, with plaintext available if your broker needs it.
• A default topic you can set once and override per block.
• Your messages are not stored. ChatMaxima connects, publishes, and disconnects.

This matters if you are on HiveMQ Cloud serverless, which has no HTTP publish API at all, so an API block could never reach it.

🧩 Publish from a flow

The MQTT Publish block takes a topic, a message body, and a quality of service level. Both the topic and the body accept variables, so the message can carry whatever the conversation collected.

• One message, the straightforward case.
• One message per item in a list, fanning out over a JSON list from a Supabase, Code or API block, with a placeholder such as {item.reference_id} usable in both the topic and the body. The whole fan-out runs over a single broker connection.
• A raw variable as the payload, if the JSON was already built upstream.

🔍 Test before you ship

The block carries the Playground and Executions tabs you already know from the API and n8n blocks, so you can publish a test message from inside the modal and read the real broker responses from past runs without leaving the builder.

✅ How to use it

1. Go to Integrations and connect your MQTT broker.
2. Open a bot in the builder and drag in the MQTT Publish block.
3. Pick the broker, set the topic and message, and test it in the Playground.

Available in all 11 dashboard languages.