> For the complete documentation index, see [llms.txt](https://en.help.firstline.cc/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://en.help.firstline.cc/master/change-log/2026.md). # 2026 {% hint style="info" %} This page is only applicable to users on the cloud version. If you are using an on-premises deployment or UAT version, some features may not be available. For full functionality, we recommend evaluating a switch to the cloud version. {% endhint %} ## Coming soon 🚀
#### **A brand-new standalone AI agent for a more flexible intelligent customer service experience** AI customer service is moving beyond simple automated replies toward**a new stage where it can independently handle tasks and connect service workflows**！ This quarter, continuing the AI agent feature in conversation scripts, we are officially launching**a brand-new standalone AI agent,**This is not just a feature upgrade, but an important breakthrough in how AI customer service operates. In the past, AI agents were one node in an automated workflow; now, businesses can**independently create, configure, and manage AI agents**and flexibly connect automated workflows and human agents according to different service scenarios, making AI a more central part of customer service. Through the integration of**business knowledge bases, AI reply instructions, automated workflows, and human agent collaboration**, businesses are no longer limited to a single-script design and can instead build a more flexible, easier-to-manage dedicated AI customer service experience based on their own needs. **By evolving AI from a workflow node into an independently operating agent, intelligent customer service gains more possibilities, including:** * **Standalone AI agent management**: centrally create and manage different AI agents, and set dedicated roles and reply rules based on service needs. * **Business knowledge integration**: specify the knowledge content the AI can reference to help handle service needs such as product inquiries and FAQs. * **Smart workflow integration**: combine existing conversation scripts and switch to specified workflows based on the conversation context, allowing AI and automated services to work together. * **AI and human collaboration**: transfer to a human agent based on configured conditions so that issues requiring human assistance can be handed over smoothly. * **External tool integration**: manage external services and tools available to AI, extending AI applications across different business scenarios. Whether it is**product inquiries, FAQ answers, service flow guidance, or handoff to a human agent**, businesses can configure AI agents according to actual needs and gradually build a more complete AI customer service workflow. #### New Custom Channels | More flexible integration for enterprise services This quarter, FIRST LINE is launching a new feature **“Custom Channels” preview version**, available to Enterprise users. In addition to existing communication channels such as LINE and Facebook Messenger, businesses can now also use APIs to connect their own websites, apps, or other service systems to FIRST LINE, further integrating customer conversations from different sources to create a customer service experience that better fits their needs. This preview version focuses on **API integration and basic conversation integration** and is suitable for businesses with technical development resources that want to further integrate their own systems with the customer service platform. With Custom Channels, businesses are no longer limited to existing communication platforms and can extend conversational services to different use cases based on actual business needs. * **Flexible integration with your own systems**: integrate business websites, apps, or other service platforms with development capabilities through APIs. * **Integrate customer service conversation flows**: allow messages from Custom Channels to connect with FIRST LINE’s existing conversation scripts and AI agents. * **Centralized channel configuration management**: create Custom Channels in FIRST LINE, obtain API credentials, and manage related settings. * **Expand service use cases**: applicable to membership services, website customer service, internal enterprise systems, and other scenarios that require conversational interaction. Whether you want to provide AI customer service in your own app or integrate your existing enterprise systems with customer service workflows, Custom Channels lets you build a more flexible service model. #### **📝 Feature improvements and enhancements** 1. Conversation filtering now includes a search by “conversation ID,” allowing you to quickly find a specific conversation by its ID, separate from the existing message keyword search. 2. The ticket filtering interface has been adjusted to integrate “Quick Options” and “Filter Templates,” centrally managing commonly used default conditions and custom filters for a more intuitive and convenient ticket filtering experience. 3. The ticket and to-do list now includes a beta “Kanban mode,” which lets you view items in columns by status and quickly change item status by dragging, making it easier to track case progress and workload distribution. 4. Improved the image upload experience with drag-and-drop upload support and clearer upload progress and status. A variety of built-in avatars have also been added, so you can quickly choose a favorite avatar without uploading your own image. 5. Timezone options have been added to personal settings, allowing time display to be adjusted based on location for easier collaboration across international teams. 6. In line with the new “AI Agent” feature, conversation scripts now include a “Return to AI Agent” node. After a specified workflow ends, the conversation can be handed back to AI for continued processing, making automated services more flexible and continuous. 7. Fixed issues with some images, data fields, and LINE menu settings to improve the stability and smoothness of script editing. 8. Improved prompts and processing flow during data export, reducing duplicate submissions and state desynchronization to make export operations more stable. 9. Improved the handling mechanism when AI replies fail, ensuring that appropriate responses can still be provided in abnormal situations and that subsequent processing flows can continue, reducing the impact on the service experience. 10. Added the Ctrl + Shift + F complete service shortcut, allowing customer service representatives to finish a service quickly without switching to mouse operations, improving daily work convenience. ## 3.21.1.1 - 10/24 1. Fixed the issue where attachments could not be uploaded continuously. ## 3.21.1.1 - 09/23 1. Fixed an issue in service desk filtering for “service tags,” where adjusting or clearing keywords could cause category display anomalies and sorting misalignment, ensuring categories are sorted correctly according to the screen settings. 2. Fixed an issue where, on some screens, the unread count indicator for internal chats might appear in the wrong position; it now displays correctly on the chat button. 3. Enhanced the handling of the staff list and online status to reduce duplicate computations and screen update overhead in environments with a large number of staff accounts, improving the smoothness of Inbox, staff menus, internal chat, and other features. 4. Fixed an issue where some long-open pages might not update immediately after staff came online, went offline, or changed status. 5. Fixed an issue in email account settings where the selected state of the SMTP encryption option might not sync correctly. 6. Improved the AI assistant’s line chart display when there is no calculable data for a specific date. The chart now maintains a more continuous and readable presentation while preserving the correct meaning of “no data for that day,” instead of incorrectly showing 0. ## 3.21.1.0
This quarter, FIRST LINE will launch multiple new AI features. As AI model capabilities continue to improve, we have also redesigned and enhanced existing features to bring more intelligent experiences and help businesses provide services more efficiently.**Building on continuous product capability upgrades, FIRST LINE is also accelerating its overseas market expansion**by continuously establishing new locations in Hong Kong, Singapore, Malaysia, Thailand, and other regions, while working with local partners to promote products and services and provide more comprehensive local support. {% hint style="info" %} This quarter’s points mechanism adds **free quota** and **dynamic points pricing**. Some features can first use the free quota provided by the system; after the free quota is exceeded, the required points will be dynamically calculated based on the actual features used and usage volume, making points consumption more aligned with actual usage. [**Click the tutorial article to learn more**](/payments-and-billing/fancy-points.md) {% endhint %} *** #### **🤖 AI Answer Agent capability upgrades** The AI Answer Agent has received multiple capability upgrades. It can now do more than just understand text entered by customers; it can also process images, recommend products, and present answers in a more intuitive way, enabling more customer service and sales guidance scenarios to be handled automatically by AI. * **Can answer from images too**\ When customers send images such as product photos or screenshots, AI will also attempt to understand the image information and respond using knowledge content, making “ask with just one image” a new service method. * **From answering questions to helping with sales guidance**\ When customers ask about products, AI can directly present recommendations in a product carousel, making the journey from asking a question to understanding and choosing a product more seamless. * **Provide quick replies to keep the conversation flowing naturally**\ AI can provide suitable quick options based on the answer content, so customers do not need to re-enter their question to move to the next step quickly. * **Even complex information can be presented clearly**\ AI responses now support table display, making product specs, plan differences, and other structured information easier to compare and read. * **The AI processing flow is visible**\ Live chat now shows AI processing status, including steps such as thinking, searching for information, and executing tools, so customers know the AI is handling the issue instead of facing an unresponsive waiting screen. * **Understand changes in customer intent faster** We have strengthened message interpretation for specific common scenarios. When customers suddenly switch questions or change topics during a conversation, AI can more quickly re-identify the actual need and reduce situations where it cannot respond in time due to context changes. In testing, accuracy improved by at least 40% in specific scenarios. * **Added exception-handling message settings** When AI cannot complete the current processing due to insufficient points, safety mechanism rejection, or other reasons, you can set a dedicated exception-handling message to reply to the customer. This is clearly differentiated from the default “no answer” reply, reducing customer misunderstanding and making message management for different failure scenarios clearer.
*** #### **AI** Assistant This quarter, we are introducing a brand-new**pilot experimental feature, AI Assistant (AI Assistant)**, trying to help you find information, keep track of work progress, and analyze service data in a more intuitive way. As long as you ask in natural language, AI Assistant can help based on the user’s existing permissions to: * **Search knowledge and SOPs**: quickly find product operation methods, policies, and related knowledge,**speeding up onboarding and daily troubleshooting**, reducing the time spent searching for information and repeatedly asking senior colleagues. * **Understand customer context**: query customer data, historical interactions, and related service records so service staff can quickly grasp the full background. * **Check pending work**: quickly find unresolved tickets and work status,**keep track of to-dos and progress**, reducing the risk of missed cases and overdue items. * **Analyze operational data**: query service volume, CSAT, and other service metrics; some analyses can generate charts directly,**lowering the barrier to data preparation so teams can identify issues and trends from data more quickly**. For example, asking “What unresolved tickets do I still have?” or “Help me summarize the CSAT trend for the last 30 days” lets AI Assistant automatically find and organize the relevant information, reducing the time spent searching across pages and manually compiling data.
{% hint style="info" %} **AI Assistant is currently a pilot experimental feature**, and we will continue expanding the data and use cases that can be queried; actual capabilities may also continue to be adjusted based on subsequent testing and user feedback. {% endhint %} *** #### **💪 AI Skills, helping businesses build reusable AI workflows** New **AI Skills**allow businesses to organize commonly used brand tone, writing rules, and workflows into reusable AI capabilities, reducing result differences caused by repeated instructions from each specialist. Administrators can create skills for different work scenarios, set usage instructions, open entry points, applicable roles, and available tools, and provide them to **AI Writer**or **AI Assistant**to use. For example, you can create skills such as “customer service replies,” “brand social media posts,” “complaint case summaries,” or “ticket quick summaries,” allowing teams to directly apply existing work rules and improve AI consistency and management efficiency. With AI Skills, businesses no longer simply let each user operate AI on their own, but can further turn**the company’s work methods and standards into manageable, shareable AI capabilities**. *** #### **✍️** AI Writer has been completely redesigned to bring AI directly into daily writing workflows **AI Writer has been completely redesigned and can now be used more easily in more work scenarios.** The new AI Writer is more deeply integrated with FIRST LINE and can be used in most**multi-line text input fields**on the platform, such as the service desk’s**message input and service notes**and others; simply select text to use AI Writer directly from the floating menu. * **AI can be used in more places**: from customer replies to internal service notes, AI can be used directly to help draft and edit,**extending AI capabilities to more frontline work scenarios**. * **Select text, modify instantly**: select existing content to polish, rewrite, or adjust the tone directly,**shortening the time from draft to usable content**. * **Without interrupting the original workflow**: AI edits happen directly in the current input field, with no need to copy-paste or switch tools,**reducing operating costs and making AI easier to become part of everyday work habits**. Making AI Writer evolve from a single feature into**a text editing assistant that fits into FIRST LINE’s daily operations**.
*** #### **📊 Customer Service Efficiency Metrics Dashboard** The dashboard now includes a beta “**customer service efficiency metrics**” view, where you can check service efficiency metrics such as first human reply, handoff, follow-up reply, and resolution time, along with**average, median, and P95**to help managers understand both general service performance and the small number of cases with longer wait times. It supports filtering by date, agent, department, team, channel, and business hours, and you can also switch between daily/weekly/monthly trends to easily observe wait-time changes across different service stages. [**View the customer service efficiency metrics tutorial**](#ke-fu-xiaolzhi-biao-yi-biao-ban) {% hint style="info" %} **This feature is currently in beta**, and the presentation of metrics and analysis dimensions may continue to be adjusted based on actual usage feedback. {% endhint %} *** #### **☎️ Phone AI real-time text (experimental feature)** For some partner companies using the new phone service, we are piloting the “**real-time text**” feature. During a call, it can display real-time text content generated after AI processing on the customer side, directly within the phone operation interface, helping agents stay in sync with the call content and reducing the burden of separately taking notes and recalling the conversation. At this stage, this feature is mainly used for functional verification and user experience testing in real service environments. Actual support scope, presentation method, and available scenarios may continue to be adjusted based on partner environments and test results. {% hint style="warning" %} **This feature is currently experimental and has only been made available to specific partner companies for now; it is not yet available to everyone.** Availability depends on the business partnership plan, phone service environment, and permission settings; the feature content and interface may continue to be adjusted based on test feedback and do not represent the final specifications of the official release. {% endhint %} *** #### **⚙️ Environment variables: centrally manage shared settings in integration workflows** Added **environment variables**management features, allowing administrators to centrally maintain values repeatedly used in Webhook and external service integrations, such as API URLs, system identifiers, API keys, or access tokens, reducing the maintenance cost of having the same settings scattered across different workflows. Environment variables support **regular variables**and**secret variables**. Regular variables can still be viewed after creation; secret variables are suitable for storing sensitive information such as API keys, tokens, and passwords. After creation, the original value cannot be viewed again and cannot be converted back to a regular variable. Through centralized management, shared settings and sensitive information can be maintained more consistently. If you change the name of an environment variable that is already in use, be sure to also check existing Webhook or integration references to avoid affecting existing workflows. *** #### **⚡**Service desk feature adjustments and upgrades This release includes multiple adjustments to the daily service desk experience and email handling logic to reduce interruptions and confusion, making it smoother for agents to handle multiple cases. * **Automatic saving of messages and email drafts**: while composing a message or email, drafts for each case will be automatically saved within the same browser session. Even if you switch to another service record, you can continue from where you left off when you return, reducing the risk of duplicate input and draft loss. * **Upgraded agent operation area integration**: integrates **AI Assistant, internal chat, notifications, phone, and message status**and other commonly used features into a shared operation area. When switching between normal pages and the service desk, most features can continue to be used and are no longer limited to a specific page. For example, while browsing different services and features, you can directly**make outgoing calls, use AI Assistant, view notifications, or communicate internally**, reducing the cost of frequent page switching and making cross-page workflows more continuous and efficient. * **Clearly separate “Start New Email” from the original conversation**: when using “Start New Email” in the service desk, a**separate email conversation**will be added, instead of being automatically merged into the original email conversation currently being replied to, while still retaining the association with the current case. Whether it is necessary to**contact internal staff separately, confirm matters across departments, or start a new side conversation**, the main and side email threads can be clearly separated, preventing different communication purposes from being mixed into the same conversation and making follow-up and record management clearer. Through this adjustment, the service desk is now more complete in**content preservation, workflow continuity, and email context management**, reducing the extra operations and interpretation costs agents face when switching work scenarios. #### **⚡ Adjustment to text template operation method** **Text templates now use a faster Slash Command (`/`) operation, allowing common replies to be integrated directly into the message writing flow.** The new version has adjusted the original text template operation method. When composing a message, simply type **`/`**to directly open and select the template you need, without leaving the content you are writing to look for the feature entry elsewhere. * **Type `/`, and quickly find a template**: call up commonly used templates directly from the input box, shortening the path to search for and apply them. * **Reduce mouse and interface switching**: templates can be selected using only the keyboard, making high-frequency replies more convenient. * **Do not interrupt the writing flow**: from input, to finding a template, to applying content, everything can be completed in the same interaction context. * **Prepare for more shortcut operations**: to support future **`@` mention**and other features, we reorganized the shortcut operations within the input box to avoid conflicts with existing operations and to give different features a more consistent usage logic. {% hint style="info" %} **Because this adjustment changes how text templates were previously operated, you may need to get used to the new workflow the first time you use it.** The new design focuses on reducing repetitive actions in high-frequency replies. Once you are familiar with the `/` shortcut method, you can complete daily replies more quickly. {% endhint %} *** #### **💬 WhatsApp embedded registration integration method** This quarter, FIRST LINE will provide a new **WhatsApp embedded registration**integration method, allowing businesses to connect WhatsApp directly from FIRST LINE and complete authorization for Business Portfolio, official WhatsApp account, and phone number according to Meta’s official process. Compared with the past, where you had to create a Meta developer app yourself and configure technical steps such as Webhook and Verify Token, embedded registration greatly simplifies the setup process and better aligns with Meta’s current recommended integration approach. * **Reduce technical setup**: no need to create a Meta developer app yourself, and no need to manually configure Webhook or Verify Token. * **Start operating directly from FIRST LINE**: the processes for authorization, account selection, and phone number selection are more centralized, reducing the operational cost of switching between FIRST LINE and multiple Meta backends. * **Easier to reuse existing WhatsApp phone numbers**: existing users can continue to use or transfer their original phone number according to the Meta screen, reducing the need to rebuild external contact methods. * **Consistent with Meta’s official process**: the interface and account setup process directly correspond to Meta’s embedded registration mechanism, making it easier to cross-reference with Meta’s official documentation later. [**View the WhatsApp embedded registration tutorial**](/feature/channels/shi-yong-embedded-signup-zheng-he.md) {% hint style="info" %} The upgrade and remapping process for existing FIRST LINE WhatsApp Cloud API accounts will still vary depending on the actual asset status. If you are already using an existing phone number, please first keep the original Business Portfolio, WhatsApp assets, and FIRST LINE Channel, then confirm the applicable process according to the tutorial. {% endhint %} #### **📝 Operation experience improvements** We are also continuously improving daily operation details: * Adjusted the rich text editor format menu to make the tool interface more centralized. * Optimized date range and time selection operations. * Improved error prompts when interface setting validation fails during data import, making issues easier to identify. * **Added message sending failure records**, when a message cannot be sent successfully due to insufficient LINE quota, a Facebook customer block, or similar reasons, the failure event will be automatically recorded and displayed in the conversation history. This allows customer service staff to directly understand why the message was not delivered, making follow-up judgments and handling easier and reducing communication gaps caused by unnoticed send failures. * **Email templates now support variables** , and “email templates” now support variables. You can use variables such as {{ contact.name }} to automatically insert customer information and corresponding content. This allows the same template to be quickly applied to different customers and scenarios, reducing repetitive editing and speeding up email reply efficiency. * When**a customer uses the “quote/reply” feature in LINE to reply to a specific message, the quoted message information will now be displayed whenever possible**, allowing agents to directly see which message the customer is replying to and reducing the time spent confirming and switching views back and forth. This feature only applies to**messages sent and triggered after the feature update**; existing messages sent before the update will not be applied retroactively. * The email channel now adds**priority ranking**settings; the sorting result will be applied to sender account selection when replying to emails, making it easier for agents to put commonly used accounts at the top, reducing the time spent searching for and selecting accounts and improving reply efficiency. #### **🛠️ Bug fixes** * Fixed an issue where some English email subjects might have text removed incorrectly. * Fixed an issue where data was not properly cleared after deselecting a custom field of the employee type. * Fixed anomalies in loading and scrolling of some live chat history messages. * Fixed an operation issue that could occur when searching with the Enter key in Chinese input methods. **※ Some features are still in testing or preparation for launch. The actual feature content and availability time are subject to the official release.** *** ## 3.9.22.5 1. Improved the display synchronization mechanism for new messages, fixing an issue in some cases where new message notifications arrived before the message data finished updating, causing the screen not to show the latest message immediately and only displaying it correctly after waiting or refreshing. ## 3.9.22.4 1. Fixed the WhatsApp contact data write issue caused by Meta’s recent data format changes. When a user has not interacted with the business number within the past 30 days, Meta may no longer provide `wa_id`. The original process relied only on `wa_id` creating contact data, causing data write failures and resulting in some messages being lost. 2. Fixed the issue where the phone blocklist could not be configured. ## 3.9.22.3 1. Fixed the statistics period filter anomaly in the “ticket status trend” section of the dashboard, and fixed the display of statistical data. 2. Fixed an issue where the “Save customer feedback” function in the bot node “AI Answer Agent” could not be saved correctly. ## 3.9.22.2 1. Adjusted the date selection input method. When selecting a date range that includes hours, minutes, and seconds, the start time will be automatically filled in after the date selection is completed **00:00:00** and end time **23:59:59**, effectively reducing operation time. If needed, you can still manually adjust the hours, minutes, and seconds. 2. Strengthened the agent handoff service flow. When an automatic message for “agent handoff service message” is set, an issue where the first message did not appear immediately on the interface because the message send time was too close has been fixed, improving the stability and consistency of message display. ## 3.9.22.1 1. Adjusted the display mechanism for the dashboard “conversation queue.” Considering long-display scenarios such as video walls, high-frequency request prompts have been removed to avoid affecting screen viewing. 2. Fixed the issue where uploaded images and videos could not display correctly in the knowledge base editor. 3. Fixed the issue where date selection could not be used properly when filtering date-type custom fields for customers, tickets, and similar items. 4. Adjusted the “maximum service count” in staff settings and restored the “no service required” option. 5. For some on-premises enterprise environments, added support for**dynamic assignment**features, which will be gradually rolled out depending on the deployment environment to improve the flexibility and efficiency of conversation assignment. 6. Fixed the email subject prefix detection mechanism to prevent normal English words such as Registration and Regression from being incorrectly identified as reply prefixes and truncated, ensuring email subjects are displayed in full and improving email recognition and search accuracy. 7. Fixed an issue where custom field “employee” types could not be saved properly after deselection; users can now clear the selected employee and save successfully, preventing the field from being forced to retain the old value. ## 3.9.22
{% hint style="info" %} ## **Live chat history contact explanation** This release will officially archive live chat contacts and conversations created before 2025 that have had no message exchange for more than one year. Due to browser mechanism limitations,**related cookies and identification information have already exceeded the retention period, and these contacts can no longer be re-established through the original live chat channel.** Therefore, most of the related conversation data has lost practical value, and it will be archived and cleaned up together to reduce unnecessary data burden. {% endhint %} In addition to continuously strengthening AI models to handle more real-world use cases, this major update also brings multiple new features that improve agent efficiency and automation capabilities through partnerships with several large multinational enterprises. Through AI, automated workflows, and integration capabilities, we help businesses create a faster, more stable, and scalable customer service and operations experience. #### AI reply inference mode launched The bot “AI Answer Agent” action node now includes a setting to “allow AI to generate a reply when no answer is found.” When the knowledge base cannot provide a clear answer, AI can offer directional suggestions, guide the next action, or supplement related information based on the question content, creating a more natural and continuous conversation experience. According to internal simulation tests, compared with fixed canned replies, enabling this feature can increase effective conversation depth by about 30% to 80% and reduce repeated questions. If applied to high-risk scenarios such as finance, healthcare, or legal services, you can choose to turn off this feature and prioritize validated content for replies to reduce the risk of AI-generated content. #### Conversation deferred processing mechanism A new “defer processing” feature has been added and can be used in the service desk. When a service cannot be handled immediately (for example, waiting for a customer or another department to reply), you can set the conversation to deferred processing and temporarily remove it from the processing list. When the set time ends or a new message is replied to, the conversation will automatically reopen, reminding the agent to continue processing. This feature helps teams manage pending items effectively, prioritize real-time needs, and avoid missing conversations that require follow-up. #### Email templates now support HTML source mode Email templates now include an “HTML source mode,” supporting direct pasting of code from third-party EDM services or existing templates. Compared with visual editing, this mode provides greater flexibility and control, allowing detailed adjustments to layout and style, and is suitable for advanced marketing or brand customization needs. Please note that different email services vary in their support for HTML, so the actual display results may differ. #### Email customer idle reminder feature The email channel now includes a “customer idle reminder” feature with the following three settings: * When the customer becomes idle * Idle reminder email content * When the customer remains idle This can help agents with follow-up tracking and reminders, reducing situations where service is interrupted because the customer has not replied for a long time. #### Automation scenarios support more triggers Automation has added more trigger points, including: * Added "On read" for customers * Conversation added "When the customer is idle" With more triggers, you can build more complex automation scenarios, such as data synchronization, idle reminders, follow-ups, or cross-system notifications.
#### Contacts page fully redesigned The "Contacts" page has been redesigned and now officially integrates contact information from the "Email" channel. You can now directly view the historical conversations and email content a specific email address has participated in, quickly review customer communication records, and improve search efficiency and service continuity. Whether for customer service follow-ups, case tracing, or cross-department collaboration, you can gain a more complete view of the customer's historical interaction context. #### Conversation queue filtering adjustments The dashboard "Conversation Queue" has added a beta filter condition. You can now filter by messages sent by customers within a recent time period, helping you quickly identify recently active conversations. At the same time, the status filter will also be redesigned as a dropdown menu, making the interface more streamlined and leaving room for future support of more status types and multilingual display. #### 💪🏼 New features and improvements 1. Phone dialing interface**Added "Call environment check" feature**Through browser-based checks of microphone, speaker, and other device statuses, it helps confirm whether the current environment supports internet calls and is used for preliminary troubleshooting of basic issues. 2. Adjusted the length limit for ticket sub-status names to improve flexibility and compatibility in English and other multilingual environments. 3. **Redesigned the custom field settings interface for the script "Prefill service record"**You can more clearly view field types and the corresponding configured data, and switch setting variables to improve flexibility and clarity of operation. 4. The live chat channel now supports, when an agent sends a video attachment and the conditions are met,**auto-converting it into a video-type message**, instead of displaying it only as a hyperlink, allowing customers to play and view the video directly in the conversation. 5. **In agent status, "Logout" will be renamed to "Offline"**to avoid semantic misunderstandings. In fact, this status does not mean the agent has logged out; it means the agent is currently not maintaining an active connection with the system, such as when the webpage is closed, the browser is asleep, the network is disconnected, or the page stops working. Therefore, using "Offline" as the status name will be closer to the actual situation; for example, when an agent closes the webpage but the login authorization is still valid, it should show as "Offline" rather than "Logout". 6. In system settings,the option for the maximum number of conversations an agent can handle has been increased from **100 to 200**, to support the higher-volume conversation handling needs of large enterprises. 7. The interface experience continues to be enhanced, including new tips and operation guides, helping new staff understand functions and workflows more quickly and lowering the learning and usage threshold. 8. The email content input limit has been increased from the original **20,000 characters** to **50,000 characters**, supporting richer and more complete email editing needs. Whether for long customer service replies, notification emails, marketing content, or complex HTML EDM templates, there is now more room for editing, reducing the need to split or simplify content due to length. 9. To meet the needs of overseas enterprise users,product pricing now supports amounts less than 1 or 0, such as **0.49, 0.99** and other price formats.This supports pricing strategies commonly used in more international markets, meeting small-payment scenarios for digital products, subscription services, and promotional activities. 10. Continuing from the previous item,the maximum character limit for product names has been increased from the original **60 characters** to **200 characters**. It supports long product name scenarios common in overseas markets, including complete information such as brand, model, specifications, capacity, color, and regional version. For example: Apple MacBook Pro 16-inch M4 Max Chip 48GB RAM 2TB SSD Space Black International Version, reducing truncation of product information and improving product recognition and management convenience. 11. For phone-number filtering in call records, based on general usage habits and query efficiency considerations,**the default will now use "prefix fuzzy search" instead of "fuzzy search"**.Since phone numbers are usually entered or recognized starting from the first few digits, prefix fuzzy search covers most query scenarios while effectively narrowing the search scope and improving query speed and user experience. 12. Legacy reporting feature**Added progress display and a cancel button**.When reports involve large amounts of data processing, you can view the current processing progress in real time and cancel an ongoing report-processing job as needed, reducing the burden caused by long waits and repeated operations. 13. **The permissions page will add explanations and tip text for some special permissions**.Through clearer purpose descriptions, restriction conditions, and scope-of-impact explanations, administrators can quickly understand the actual effect of each permission, reducing misconfiguration and misunderstandings. 14. The ticket follow feature has been changed from**a pop-up prompt to a dropdown menu**, improving display and operational compatibility on some tablet devices. 15. The Instagram channel has**added support for the "Share Post" message type**, allowing customer-shared post content to be displayed correctly and improving message completeness and conversation experience. 16. For contacts in the email channel,**prompt messages will no longer be shown or new service creation blocked just because there are other ongoing services**. The usage scenario for the Email channel is different from live chat. The same contact can exist simultaneously in multiple independent email topics, recipient mailboxes, or handling workflows. For example, the same contact can send a "return" topic to mailbox A for handling, and also send a "purchase" topic to mailbox B for handling; the two are different business scenarios and do not affect each other. Therefore, the relevant logic has been adjusted so the Email channel can create and handle multiple independent services at the same time according to the actual usage scenario. 17. Fixed an issue where when the system setting "Password history duplication restriction" was set to not check, it would still trigger password history duplication validation. 18. Fixed an issue where custom field names ending with "\*" caused the regular expression rule to fail. 19. Fixed the filtering logic for "Service Level" on the dashboard. Because the phone channel does not have channel account settings, data could not be displayed properly under certain filter conditions. After this adjustment, statistics and filtering logic for the phone channel will be handled correctly, avoiding display anomalies caused by missing channel account information. 20. Fixed an issue where "online channels" could not properly search conversations in the service list and contacts page. You can now correctly search online channel conversations on the relevant pages, improving data lookup and service tracking efficiency. 21. Fixed an issue where using the service desk "Start a new email" feature could prevent emails from being sent normally if the recipient was a newly created email contact at that moment. 22. Fixed an issue where when using the email channel "Send test email" feature, if the test failed due to incorrect account password, abnormal SMTP server settings, connection failure, or other integration issues,**only a generic `Server Error` message**was shown, making it impossible for users to determine the actual cause. 23. For some users using the new phone service (R8), the call status display has been enhanced with more explicit status information, such as "Logout: connection timed out" and "Logout: duplicate login," making the cause of the issue more intuitive and easier to understand. 24. For some users using the new phone service (R8), fixed an issue where no audio could occur in certain transfer scenarios. For example: after a call is answered by agent A and transferred to internal extension B, under certain conditions the call may lose audio after some time. 25. Finally, some interface text, explanatory content, and prompt messages will be adjusted to avoid being overly technical, insufficiently intuitive, or inconsistent in perspective. ## 3.6.10.4 - 04/18 #### Explanation of changes to the conversation-initiated contact process We will redesign the service process for agents proactively initiating contact. In the future, the conversation record will be assigned based on the location of the operation to which service it belongs, instead of automatically linking to the earliest unfinished service. This adjustment can prevent different service topics from being mixed into the same record, reduce the risk of data confusion, and improve the accuracy of service records. **The revised rules are as follows:** 1. **Initiate contact inside the service desk**\ The conversation record will be linked to the service currently being viewed. 2. **Initiate contact outside the service desk**（e.g. customer page, single service page, etc.）\ The system will automatically create a new service record. This item will be launched first on some channels (such as LINE and email), with other channels (such as SMS) expected to be updated gradually within this quarter. {% hint style="info" %} In the original design, once a customer created a "return" service, as long as the service had not ended, any subsequent contact for any reason would continue to accumulate in the same service, and no new service would be created separately. This could easily mix conversations from different service topics together, making it difficult to determine the actual handling content when querying or analyzing later. In the latest design, agents can decide based on the actual scenario whether to reply within an existing service or start a new contact separately, so that services with different topics can be clearly distinguished, improving data readability and analytical accuracy. {% endhint %} #### 💪🏼 New features and improvements * **Dashboard "Service Level" channel filter statistics anomaly,**Fixed an issue where filtered data was displayed incorrectly. * **Custom field hierarchical menu operation anomaly,**Fixed an issue where an unnecessary warning window appeared when reopening batch editing. ## 3.6.10.3 - 04/10 1. An Instagram channel has been added to support displaying post messages sent by customers, making the content more complete and easier to review and handle later. 2. An issue where points were displayed incorrectly in the subscription points plan has been fixed. ## 3.6.10.2 - 04/01 1. Fixed an issue in the chatbot "Business Hours" where the "Specific Date" dropdown under special time periods did not display dates. 2. Adjusted how the points subscription plan is displayed. 3. Adjusted an issue where after modifying a customer's name in the service desk, some related interfaces did not sync the update immediately. 4. Adjusted the issue where the "&" character in messages containing hyperlinks was escaped and displayed. 5. Fixed an issue where the simulated bot feature in scripts could not run properly at some nodes. 6. **Fixed the translation error in the IP allowlist,**The issue where "System" and "API" were displayed in reverse on a single page has been fixed. 7. **Fixed the ticket email-sending alert issue,**The abnormal alert triggered when using "Send Email" in a ticket has been fixed. ## 3.6.10.1
#### FANCY points monthly subscription plan launched,**more effectively control AI usage costs** This time we officially launch the **FANCY points monthly subscription plan**, helping enterprises manage AI usage costs more effectively. This plan is mainly designed for medium and large enterprises and is divided into low, medium, and high tiers. Enterprises can obtain a certain number of points through a fixed monthly fee and use AI and other advanced applications at a more favorable unit cost. At the same time, it can reduce the complexity of cross-department billing processes and internal cost allocation, making budget management clearer and more predictable. ***We hope to help enterprises truly adopt and implement AI with transparent, reasonable, and predictable pricing, continuously gain new features, avoid the black-box nature and cost overruns of outsourced projects, and let enterprises invest resources in actual application and results rather than repeated coordination and uncertainty risks.*** #### New Persona architecture launched, improving AI conversation understanding FIRST LINE AI now introduces a newly designed **Persona architecture**, enabling more accurate understanding of customers' follow-up questions, reducing response gaps caused by incoherent context, and improving overall conversation quality. In addition, we also plan to launch more powerful and accurate AI marketing and product recommendation features, further strengthening conversion and interaction experiences in scenarios such as e-commerce. #### Dashboard adds "Service Tag Trends" chart The dashboard now includes a beta “**Service Tag Trends**chart, used to view changes in various service tags within a specific time period. By observing rising or falling tag counts, managers can identify potential service factors. For example, if one tag continues to rise, it may indicate that related issues or needs are increasing and should be addressed early. #### Dashboard adds "Customer Tag Trends" chart Dashboard adds a "Customer Tag Trends" chart, used to view, within a specific time period, statistics by tag assignment time and analyze changes and trends in service tags. By observing changes in tag counts, it can help determine the actual situation; for example, if a product-related tag continues to rise, it may indicate increased demand or concentrated issues for that product. {% hint style="info" %} ## "Service Reason" renamed to "Service Tag" 🏷️ To align terminology consistency across regions and countries, and to support the expansion of related features in the future, "**Service reason**" will be renamed to the more straightforward and intuitive "**service tag**" to lower the learning curve and make it easier for new staff and cross-regional teams to get started and use. {% endhint %}
Two**beta features** have been added to the conversation feature:**Message recall**and**Contact blocking**to help agents manage workflows more effectively when handling conversations and create a smoother service experience. #### Live chat adds message recall feature In the live chat channel, you can now**recall a single message**. When an agent sends a message by mistake, or sends the wrong link or file,**within a limited time**you can recall that message, reducing communication and service disputes caused by mis-sends. #### Contact blocking feature to prevent harassment messages This launch introduces **Contact Blocking (Beta)** feature. You can block**a single contact**.After blocking, messages sent by that contact on that channel will**no longer trigger bot workflows or auto-replies**and similar actions, but they will still be received and displayed normally. This feature can reduce interruptions caused by spam or harassing messages, eliminating the need to handle blocking scenarios through other methods and allowing agents to focus more on normal service. #### Service record reopening mechanism, improving SLA and process accuracy In order to provide more accurate **SLA and total service time** data in the future, and support more extended features (for example, creating automation workflows through APIs), the handling of service records will be adjusted. Service records, after being**completed, will no longer allow direct modification**, and will instead adopt a more standard**reopen** **(Reopen) mechanism**. When a service is reopened: * the service will return to**the original assigned agent** * the status will change back to **in service** to establish a more standardized service workflow structure for future APIs and automation processes. #### 💪🏼 New features and improvements 1. The hierarchical menu for custom fields will be separated into a dedicated management item. In the future, you can first design a hierarchical menu classification and reuse it across different custom fields, making the data structure more consistent, management more efficient, and future applications easier to expand. 2. Orders and items will add tax detail data, allowing agents to clearly break down the tax composition, shorten troubleshooting time, and make finance and operations audit processes more consistent. 3. Adjust the message data structure to strengthen read-message efficiency in large**data-volume scenarios**, improving overall speed and operational smoothness. 4. Redesigned the "Order" view screen, splitting information into multiple clear sections to better fit e-commerce use cases and improve overall reading and lookup efficiency. 5. Product pricing now supports decimal input to accommodate amount formats and standards in different countries and regions. 6. **Removed the mechanism in the service desk message input box that parsed `{{****}}` variables only upon sending**.Variables are now parsed as soon as a "message template" is selected, so agents can see the full content before sending and directly edit and adjust it before replying. 7. **Removed the four-digit verification code mechanism when deleting operations**, allowing users to complete tasks more quickly when handling multiple records. 8. Adjusted the character limit for custom field names and options, especially optimized for English-speaking country use cases, to avoid display truncation or inability to fully enter longer words. 9. Added a "Shipping Method" field to order data {% hint style="info" %} ## **Live chat history contact explanation** Planned for **live chat contacts created before 2024 and with no message exchanges for over one year** to be**organized**. Due to browser characteristics, the related cookie and localStorage identifiers have long exceeded their retention period. These live chat contacts can no longer be contacted through the original channel, and their contact conversations have also lost practical value, so they will be removed together. {% endhint %} ## 3.0.0.7 - 03/08 * Fixed the dashboard "Script Event Tracking" dropdown showing only **50 options** problem. * Adjusted the validation logic for "Allow duplicate phone numbers" in system settings. When agents do not have permission to view all customer data, the original mechanism could cause duplicate entries to still be created even when duplicate numbers were not enabled. ## 3.0.0.6 - 02/25 * Fixed the channel category keyword filtering feature ## 3.0.0.5 - 02/08 * Fixed an issue where when performing batch operations on customers, if the filter conditions involved custom tables, some customers might be missed during the operation (for example, when applying tags). * Fixed the issue where change logs for customers and tickets could not be displayed, ensuring related changes can be viewed and tracked normally. * The crawler preference list has been adjusted to avoid cases where carousel messages on the Instagram channel could not be displayed because Meta's special crawling logic was triggered by the first image possibly being a cover image. ## 3.0.0.4 - 01/30 * Adjusted the display status of historical messages during loading to prevent users from mistaking temporary loading for blank or abnormal content. * Fixed the issue where the play button in the phone satisfaction report could not be used. * Adjusted the email channel creation process by adding corresponding validation mechanisms for different authentication methods to avoid incomplete settings or authentication errors. * Adjusted the dependency relationship between "Password history duplication restriction" and "Password reuse restriction"; the two settings can now be enabled and managed independently. * Adjusted the timestamp display of event records in conversation messages to help agents more clearly understand the conversation timeline and related workflow. ## 3.0.0.3 - 01/14 * Adjusted a specific case where the bot script name is input **50-character limit**, the validation and prompt mechanism for saving names that are too long. * Added order-related status options**multilingual display support**. * Fixed**Service reason settings cannot be sorted** problem. * Indonesian and Filipino have been added to the AI translation service, helping multinational enterprises reduce language communication costs. * Fixed the issue where certain operations in the customer management page could cause customers to be added twice. * In response to the built-in **Chrome and Safari**) [**Autoplay Policy**](https://developer.chrome.com/blog/autoplay) restriction, when the user has not performed any interaction (such as clicking), the notification sound cannot play automatically. This update will add**prompt explanation messages**, helping agents understand and attempt to lift the restriction.
## 3.0.0.2 - 01/13 * Fixed, under certain conditions,**message ordering anomalies and duplicate display** problem. * Fixed the latest message**timestamp not displayed** problem. * Fixed**The unread marker display for automatic event reply messages is out of sync**issue. * For some users, due to**unstable network connections**, a desynchronization between the assignment and service screens could occur. A compensation mechanism has been added. When the network recovers within a certain period, the system will**automatically attempt to refresh the data**, reducing the need for manual intervention; at the same time, it will also try**to add additional records**, to facilitate later issue diagnosis and tracking.