Krypton

Last update November 13, 2026

AI Agents

AI disclosure message [#ai-disclosure-ai-agents]

The new AI disclosure message setting automatically shows end users a message that tells them they are interacting with AI. The setting is on by default in AI agents, chatbots, and journey elements, and shows the disclosure at the start of the conversation. A new page gives general guidance, not legal advice, on the transparency requirements of Article 50 of the EU AI Act(opens in new tab), including when to disclose and how to write a clear disclosure message.

Key details:

• AI agents: Go to AI Agents → My agents, open the agent, and find AI disclosure message on the Agent tab. The message appears at the start of the first message from the agent.
• Chatbots: Find AI disclosure message on the Settings tab of the chatbot. If the chatbot uses an AI agent and both have the setting on, the chatbot setting takes priority.
• Journeys: Find the setting in the send elements for SMS, Viber, Messenger, Email, Mobile Push, Web Push, RCS, WhatsApp, LINE, MMS, Apple Messages for Business, In-app, and Zalo. In the Transfer call to AI agent element, the setting is available only when you select Speech to speech.
• Turning the setting off: If you turn off the setting where AI talks to end users directly, your AI agent, chatbot, or journey might not meet the transparency requirements.

For more information, see AI transparency and disclosure, Configure your agent, AI disclosure in chatbot settings, AI disclosure in Journeys, and AI disclosure in Transfer call to AI agent.

Customer Profiles

Customer context [#customer-context-customer-profiles]

Customer context turns conversation history into a summary and facts about each customer. When a chatbot, journey, or AI agent hands off a conversation to a human agent, the agent gets a briefing and can help right away, so customers do not have to repeat themselves. AI analysis captures a summary, facts, and the customer's sentiment from each analyzed conversation.

Key details:

• Custom facts: Define the details that matter to your business, such as an order number or a policy number, on the Custom facts tab in Customer Profiles → Configuration. Use the Enabled toggle on the same tab to turn Customer context on or off. Select Suggest facts to get fact suggestions based on your past conversations, and approve the ones you want to keep.
• Person profile: The Context section shows a summary across the person's conversations and the most recent value of each fact. Memory history lists each analyzed conversation with its facts and summary.
• Inbox: The AI customer context card in My Work gives agents a briefing at handoff with the current interaction, a customer summary, and the customer's past interactions. To show the card, turn on Show AI customer context to agents in Inbox → Settings → Productivity → Inbox AI → AI customer context tab, and select the queues.
• Insights: Customer context requires AI analysis to be turned on in Insights settings. Customer context adds to your AI usage.
• Context API: Manage custom facts, get the summary and the memories of a person's past conversations, search memories, and receive webhook events when the context changes.
• Context MCP server: Give AI agents access to Customer context, including the person summary, past conversations, memory search, and custom facts.

For more information, see Customer context, Custom facts, Context on the person profile, AI customer context in Inbox, and AI customer context settings. For the API and MCP, see the Context API reference and Context MCP server.

Inbox (Conversations)

Inbox AI [#inbox-ai-inbox]

Inbox AI is now part of Ask AI, the Infobip AI assistant. It replaces the earlier AI assistant and Inbox copilot naming and brings all AI features that help agents reply faster and resolve conversations into one place. When a conversation is open, Ask AI works with the context of that conversation, so agents can ask questions about it, find information in the knowledge base, and get suggested replies.

Inbox AI includes the following features:

• AI conversation summary: Generate a short summary of the conversation to quickly understand the context.
• Enhance writing: Fix grammar and spelling, and rephrase messages.
• Translate agent messages: Translate replies into the customer's language before sending.
• Suggested replies: Get reply suggestions based on the knowledge base and previous conversations.
• Fill in forms with AI: Fill in form fields from the conversation history. This feature is charged per use.

Key details:

• To open Ask AI, open a conversation in My Work and select the AI icon in the page header. Select a suggested prompt, for example, Suggest me a reply, Find similar conversation, or Summarize previous conversation, or enter any question in the Ask me anything field.
• The AI icon in the message editor toolbar gives quick access to Generate summary, Fix grammar and spelling, Rephrase, Translate, and Suggest a reply.
• A summary generated in Ask AI appears in a Conversation summary card. Agents can edit it, regenerate it, save it to the Summary field with Add as summary, or save it as an internal note with Add as note. Agents can also select Generate summary with AI in the Conversation context card.
• A suggested reply can include sources from the knowledge base or similar closed conversations. Select Add to composer to add the reply to the message editor and edit it before sending.
• Each conversation has its own Ask AI session, and context is not carried over between conversations. Sessions are retained for 15 days.
• Administrators configure Inbox AI in Inbox → Settings → Productivity → Inbox AI. Use the toggle next to each feature to turn it on or off, and use Edit queue access to apply it to all current and future queues or to specific queues.
• Suggested replies are turned on per queue. Optionally, add a knowledge agent to a queue to use its knowledge base. Without a knowledge agent, suggestions are based only on previous conversations.
• For Fill in forms with AI, choose the form, the fields, and the method: On demand or Automatic form filling. Only the conversation history is provided to the AI.

For more information, see Inbox AI in Inbox and Inbox AI settings.

Broadcast history in Inbox is now generally available [#broadcast-history-ga-inbox]

Broadcast history is now generally available in Inbox. It was released as Early Access in the October release. Agents can see messages that were sent to a customer through Broadcast directly in the Inbox conversation thread. When a customer replies to a Broadcast message, agents see the original message and understand the context of the reply without leaving Inbox.

Key details:

• Turn it on in Inbox → Settings → Basic settings → Account preferences → Broadcast history. The setting applies to the entire account and is off by default.
• A Broadcast message is displayed if the conversation uses the same sender, the Broadcast was sent within the last 15 days, and it was sent through SMS, RCS, WhatsApp, Viber Business Messages, or Viber Bot.
• Broadcast messages appear as outbound messages with the Broadcast label. Hover over the sender number to see the channel and sender.
• Broadcast messages do not start the conversation and do not affect analytics or SLA calculations, but they are included in conversation transcripts.
• Managing the setting requires the inbox/conversations/Broadcast-history/manage permission, included in the Inbox setup role by default.
• Available only in Inbox.

For more information, see Broadcast history and Message features in My Work.

Meta Business Agent handoff in Inbox and Conversations [#meta-business-agent-handoff-whatsapp-inbox-and-conversations]

Meta Business Agent is an AI agent built into the WhatsApp Business Platform that answers customers directly. If you use Meta Business Agent on your WhatsApp sender, Inbox and Conversations now show its conversations so that supervisors can follow them and step in when a customer needs a human agent.

Key details:

• Conversations that Meta Business Agent is handling are shown in All work → Automated conversations, which is available only to supervisors. The Agent column shows the name of the Meta Business Agent, and the Agent filter narrows the list to a specific one.
• Supervisors can open these conversations and follow them in real time. While Meta Business Agent is in control, no one can reply in the conversation.
• Control moves to Inbox through a handoff, when Meta Business Agent escalates the conversation, or a takeover, when a supervisor selects Assign to me or assigns the conversation to an agent before it is escalated.
• After a handoff, the conversation is routed in the same way as any other conversation, based on your routes and automatic agent assignment settings. Until an agent is assigned, it is shown in All work → Unassigned.
• When the agent closes the conversation, control returns to Meta Business Agent, which answers the customer the next time the customer sends a message.
• If Meta Business Agent does not transfer control to Inbox, the conversation is closed after 24 hours. When the customer starts a new session with Meta Business Agent, a new conversation is opened.

For more information, see Meta Business Agent handoff in Inbox and Meta Business Agent handoff in Conversations.

RCS rich content in Inbox and Conversations [#rcs-rich-content-inbox-and-conversations]

RCS in Inbox and Conversations is no longer limited to text messages. Agents can now send and view rich RCS content, such as files, locations, suggested actions, rich cards, and carousels.

Key details:

• RCS templates: Create RCS templates in Settings → Productivity → Templates so agents can use them in ongoing RCS conversations. Supported content types are Text, File, Rich card, and Carousel (up to 10 cards). You can add suggested actions to every content type: Reply, Open URL, Dial a phone, Show location, Request location, and Create calendar event.
• Request a location: Agents can ask a customer to share their location by selecting Request a location in the message composer toolbar. The customer's response is shown as a location message in the conversation.
• Voice messages: Agents can now record and send voice messages over RCS, in addition to WhatsApp.
• Rich cards and carousels in the conversation: Rich cards and carousels sent to the customer, for example, by a chatbot or from a template, are shown the same way the customer sees them. Rich card buttons and suggested replies are for display only, and agents cannot select them.
• Customer replies: When a customer selects a reply button or suggested reply, the conversation shows an inbound message with Pressed: followed by the button label. Contacts and GIFs that customers send are shown as files.
• To share a location over RCS, agents must write a message in the composer first. RCS does not support sending a location without text.

For more information, see RCS in Inbox, RCS templates in Inbox, RCS in Conversations, and RCS templates in Conversations.

WhatsApp media carousel messages in Inbox and Conversations [#whatsapp-media-carousel-inbox-and-conversations]

When a chatbot or a Journeys flow sends a media carousel message to a customer, Inbox and Conversations now show it as a carousel in the conversation history. This helps agents understand what the customer saw before the conversation was transferred to them. Agents can also send media carousel messages during ongoing conversations by using WhatsApp templates with rich content.

Key details:

• Each card shows the media header (image or video), the body text, and the buttons. Scroll horizontally to see all cards.
• When the customer selects a quick reply button, the conversation shows the card that the customer replied to, followed by Pressed: and the button label.
• Media carousel is now a supported content type when you create WhatsApp templates with rich content in Settings → Productivity → Templates.

For more information, see Media carousel messages in Inbox, WhatsApp templates with rich content, and Media carousel messages in Conversations.

Generative voices in call waiting strategies [#generative-voices-call-waiting-strategy-inbox-and-conversations]

When you set up a waiting strategy message for calls, you can now choose generative voices in addition to standard and neural voices. Generative voices are marked (generative) in the Voice list and sound the most natural and human-like. Use them to keep the same natural voice throughout the call, from the IVR or Voice AI agent to the waiting strategy.

Key details:

• Neural and generative voices have additional charges, and generative voices cost more than neural voices. If neural or generative voices are not available for your account, contact Support or your account manager.
• To announce the position in the queue, check Available placeholders under each voice in the Voice list. Some generative voices do not support the Ordinal and Cardinal placeholders.
• With Adaptive numbering, generative voices always use cardinal numbers (one, two, three), not ordinal numbers (first, second, third). Write the message so that it works with cardinal numbers, for example, "Your position in the queue is number two."

For more information, see Waiting message language and voice selection in Inbox and Waiting message language and voice selection in Conversations.

SLA for outbound conversations [#sla-outbound-conversations-inbox-and-conversations]

SLAs can now apply to outbound conversations, which are conversations that an agent starts, for example, by sending a template message. On some channels, such as WhatsApp, the agent cannot send further messages until the customer replies and opens the 24-hour customer care window. Because of this, SLA tracking for these conversations starts only when the customer engages.

Key details:

• When you create an SLA, use the new This SLA applies to conversations started by setting to select Customer, Agent, or Customer or agent. SLAs created before this setting was available apply only to conversations started by customers, and the setting does not change how SLAs track those conversations.
• For conversations started by an agent, sending the initial message does not start SLA tracking. Tracking starts with the customer's first reply, and Inbox selects the matching SLA policy at that point.
• First response time is the time from the customer's first reply to the agent's first message after it. Total resolution time is the time from the customer's first reply to the status being set to Solved. Next reply time works the same way as for conversations started by customers.
• If the customer never replies, the conversation is not tracked and cannot breach the SLA.
• When the queue or the customer of an agent-started conversation changes, the SLA calculation always begins from the customer's first reply.
• Broadcast messages are still excluded from SLA calculations.

For more information, see Outbound conversations in Inbox and Outbound conversations in Conversations.

Live analytics [#live-analytics-inbox-and-conversations]

Live analytics now gives supervisors more visibility into agent activity and more control over how the dashboard looks.

Human agents table improvements

The Human agents table in Live analytics now gives contact center supervisors and team leads real-time visibility into what each agent is handling, so they can intervene faster and manage the floor more effectively.

What is new:

• Conversation statuses: See the number of conversations in each status for every agent.
• Current workload: See how many active conversations each agent is handling by channel, compared to their maximum workload.
• Capacity rate: See how much of their capacity each agent is using, as a percentage.

Live analytics customization

You can now customize Live analytics to focus on the information that matters most to you.

What is new:

• Columns: Select which columns are shown, change their order, and resize them.
• Widgets: Control which widgets are available on the dashboard and change their order.

For more information, see Live analytics in Inbox and Live analytics in Conversations.

RCS

10 second minimum message validity [#min-validity-rcs]

RCS messages now enforce a minimum validity period of 10 seconds. If you send a message with a validity period shorter than 10 seconds, the platform overrides it to 10 seconds to account for queuing and processing time.

Previously, very short validity periods (6 seconds and below) caused delivery inconsistencies with Google, which implemented its own override mechanism at 6 seconds.

Key details:

• No breaking change. You do not need to update your integrations.
• Validity periods of 10 seconds or longer are not affected.
• If you use failover channels, rely on the RCS delivery report to determine when to send the failover message, not on the validity period defined in the original message.

For more information, see Additional features and Send RCS message.

Media payload size in delivery reports [#media-payload-size-rcs]

RCS delivery reports now include the total size of all media assets in the message, both the original size measured on the Infobip platform and the compressed size reported by Google. This helps you evaluate potential additional charges for media payloads above carrier thresholds.

Previously, there was no visibility into how media compression affected the payload size of RCS messages.

Key details:

• The original media size (measured by Infobip) and the compressed media size (reported by Google) are both included in the delivery report.
• Applies to media sent in rich cards, file messages (image, video, audio, or file), and carousel messages with multiple media links.
• Certain US carriers charge for media payloads above 5 MB. Google compresses media before delivering it to handsets.
• No billing changes. This is a reporting enhancement only.

For more information, see Receive RCS delivery reports.

Reporting

Live Chat and Open Channel in Financial reports, AgentOS reports, and the Billing Usage API [#live-chat-open-channel-reporting]

Financial reports, AgentOS reports, and the Billing Usage API now support Live Chat and Open Channel. Review usage for these channels alongside your other channels, reconcile channel and AgentOS processing costs against your invoice, and use the data in your own reporting workflows.

Key details:

• AgentOS reports: Show Live Chat and Open Channel usage and the AgentOS processing cost that this usage triggers.
• Financial reports: Select LiveChat and Open Channel in Channels and services.
• Billing Usage API: Filter and group data with the LIVE_CHAT and OPEN_CHANNEL categories.

For more information, see Financial reports, AgentOS reports, and the Billing Usage API. For category codes, see Categories in the Billing data reference.

Resources and numbers

Automated document prefilling for alphanumeric senders [#document-prefilling-resources-and-numbers]

Many countries require country-specific documents, such as registration forms and letters of authorization, to register an alphanumeric sender. Previously, you filled in each document yourself, sometimes 5 to 10 documents per country. With automated document prefilling, you provide the information once in the request form, and Infobip fills in the required documents in the correct format.

Key details:

• Supported countries: Prefilling is available in more than 60 countries.
• Your part: You sign, stamp, or add a company letterhead only when a document requires it. Infobip returns the prefilled document, and you download it, complete the action, and upload it again.
• Other documents: You still upload documents that the country requires from you, such as an ID document or a company registration certificate.
• Web interface: Prefilling is on by default for supported countries.
• Resource Request API: Opt in per request with featureOptIns.documentPrefill set to true.
• Provision MCP: Prefilling is on by default. The agent asks for the extra information during the conversation.

For more information, see Automated document prefilling, Document prefilling in the API guide, and Sign and upload prefilled documents.

Port US numbers in and out for Voice [#us-number-porting-resources-and-numbers]

You can now port US numbers to and from Infobip for Voice directly in the web interface. Port in numbers from your current carrier to keep using them on Infobip, or port out numbers from Infobip to another carrier.

To start, go to Channels and Numbers > Numbers > My numbers, select Bring your own number, and then select Port in 10DLC (Voice, US) or Port out 10DLC (Voice, US).

Key details:

• Supported numbers: US virtual long numbers.
• Numbers per request: Enter up to 100 numbers, or upload them from a CSV file.
• Port in: Enter the details from your current carrier account and upload a Letter of Authorization (LOA) signed by the account holder. Optionally, set the E911 emergency address and the CNAM caller ID for your numbers before or after you submit.
• Port out: Enter the port-out details and the new network service provider that receives the numbers.

The Import phone numbers page is now Bring your own number, and covers number imports from other providers as well as US porting.

For more information, see Port in US numbers and Port out US numbers.

Subscriptions

Instagram events in webhook subscriptions [#instagram-webhook-subscriptions]

Instagram is now available as a channel in webhook subscriptions, in the web interface and the Subscriptions API. Previously, to receive Instagram delivery reports, you had to add a webhook URL to every send request or use an account-level URL. Now you create one subscription for Instagram and receive the events at your webhook.

Key details:

• Available events: Delivery (DELIVERY) and Click (CLICK). Use click events to see which messages get clicks, and on which link.
• Setup: In the web interface, select Instagram as the channel when you create a subscription, or create the subscription with the Subscriptions API.
• Subscription groups: Instagram works with subscription groups, so you can send the same event to more than one webhook.

For more information, see Manage webhook subscriptions and Available webhook events.

Viber

Transactional template duplication [#template-duplication-viber]

You can now duplicate transactional templates for Viber Business Messages directly from the web interface. Use this to create a similar template based on an existing one or to edit and resubmit content from a rejected template without starting from scratch.

Previously, creating a similar template or resubmitting a rejected template required filling in all fields manually.

Key details:

• Select Duplicate template on any existing transactional template. The content is pre-filled, and you can modify it before submitting.
• Available through the Infobip web interface only.

For more information, see Viber message templates and Transactional templates.

Voice and Video

Ringless Voicemail [#ringless-voicemail-voice-and-video]

Ringless Voicemail is a new voice channel that delivers a text or pre-recorded voice message directly into the end user's voicemail box without ringing their phone. Use it as a standalone channel, as failover for SMS, or as an alternative to SMS through Smart Messaging.

Key details:

• No interruptions: The end user's phone does not ring. They receive a standard voicemail notification and listen to the message at their convenience.
• Standalone or failover: Send messages on their own through the API, or use Ringless Voicemail as failover when SMS delivery fails.
• Smart Messaging alternative: Ringless Voicemail can be set up as an alternative to SMS as part of the Smart Messaging solution. Contact your Infobip Account Manager for this option.
• Availability: United States and Canada only.

For more information, see Ringless Voicemail and Send ringless voicemail.