Changelog

​

Voice

• Voice agents are available in beta. A voice agent answers and places phone calls, takes calls from your website, runs tools, searches knowledge bases and hands calls to people. Voice is on paid plans once Tars turns it on for your organization. See Voice agents.

​

Tools

• The toolkit directory adds more than 600 toolkits, for more than 1,400 in total. Search by provider name. See Connect from the Directory.

​

Agent builder

• On paid plans, the Model picker of an AI Agent gambit adds Meta models, such as Llama 4 Scout and Muse Spark. See Model.
• A Button gambit answer now holds both the value and the label of the choice, as .value and .label. See Read the selected value and label.
• After a Button gambit, an AI Agent gambit with Auto-Greet answers the option the end user selected instead of a general greeting. See Behavior.

​

Campaigns

• A paused campaign that has not finished preparing recipients keeps its segment locked. The error names the campaign, so you know which one to resume. See The segment is locked while recipients are prepared.

​

End users

• When Tars merges an anonymous profile into an identified one, segment membership and segment data move with it, even while a campaign uses the segment. See How profiles merge.

​

Knowledge

• A retriever’s media now always uses the embedding model of its text index. The media Embedding model field is read-only, and changing the text embedding model moves the retriever’s media to the new model. See Configure a retriever.
• A retriever saved with its own media embedding model, or with the original media index, keeps working. It moves to the text embedding model the next time you save a change to its text index or media settings, and its images are embedded again.
• A retriever that uses the same media model as the knowledge base’s shared media index no longer stores a second copy of every image.

​

Agent builder

• With Allow Skip on, an end user can skip a Media Upload gambit again. A failed upload no longer skips the step. See Media Upload gambit.
• A Star Rating gambit shows the number of stars you set, and the answer shows against that number instead of 5. See Star Rating gambit.
• A long prompt no longer overflows the instructions dialog of the AI Agent gambit. See AI Agent gambit.

​

Tools

• Sort in the toolkit directory adds Recently added. See The Directory tab.
• A retired toolkit leaves the directory. Its existing connections keep working, and you can still manage their actions or delete them. You cannot add, reconnect, or change the account of a connection to it. See When a toolkit is retired.

​

Voice

• A call’s outcome is now a category, such as Resolved by the agent, Transferred or No conversation. See Outcomes.
• Call exports show each call’s outcome category and an Outcome reason. See Export calls.
• The End Call tool now ends a call only when the caller says goodbye or asks to end. Your prompt can also tell the voice agent to end once the request is handled. See Built-in tools.
• Holding lines take turns while the caller waits, and none plays twice. You can add up to ten lines per language. See Configure the settings.
• Version 5 of the Handbook preset text stops the Professional + Conversational tone from ending calls early. An existing voice agent keeps its version until you select Update preset text and publish. See Update the Handbook preset text.
• A flow node no longer tells the caller that the voice agent cannot do a task that a later flow node does. Publish an existing flow again to get this. See What a flow node knows about later flow nodes.
• On the web call screen, live captions stay where you scroll. A button shows the number of new lines and takes you back to the latest caption. See Change how the call screen looks.

​

Voice calls API and webhooks

• A call’s outcome is now a category, such as resolved_by_agent or transferred, instead of true or false. The analysis model’s own yes or no is in analysis.successful. See Outcome and analysis.
• Voice webhook events add outcomeReason to a call’s analysis, and ending.by says when a member ended the call. See Webhook events.

​

Analytics

• The Labels card shows each label’s conversations, messages, and response times. Label changes always reach analytics, and the Use labels in analytics switch is gone. See Labels.

​

Knowledge

• The Free plan stores up to 1,000 chunks across all knowledge bases. The Chunk limit button on the Knowledge bases page shows your usage. At the limit, your agents keep answering. See The chunk limit.
• Knowledge base images now show in the Image Gallery and Cards components. See Component gambit.

​

Agent builder

• The AI Agent keeps its internal working out of the replies end users see. See AI Agent gambit.
• Migrate to components moves the legacy output and input components of an AI Agent to a Component gambit. See Legacy capabilities.

​

Live chat

• A label you tick on a conversation opened from a link now saves. See Use the Agent Workspace.

​

Developer

• The MCP analytics tools accept date-time windows for reply_breakdown, human_report and whatsapp_messages. Wrong parameters return invalid_input with the reason. See Date formats.

​

Deploy

• Auto-open widget can target All devices, Mobile, or Desktop, and waits for a Delay you set. It does not open again after the end user dismissed it in that tab. See Auto-open the widget.

​

Knowledge

• Large knowledge bases index faster. See Configure a retriever.

​

Analytics

• The WhatsApp messages card counts replies sent on WhatsApp, the replies Meta billed, and each number’s free allowance. Add it with Add chart. See WhatsApp messages.

​

Developer

• AI tools connected over MCP can read your analytics with five read-only tools. See Connect AI tools with MCP.

​

Analytics

• The AI performance card shows how conversations ended and who handled them, next to how fast the AI replies. Reply time details opens the AI replies drawer. See AI performance.
• The Team card shows handoff outcomes and each human agent’s conversations, response times, and CSAT. See Team.
• Key metrics adds the CSAT tile with its drawer, and the live Waiting for an agent and Open now tiles. The CSAT, Handled by, and Reply time cards are removed. See The Key metrics gallery.
• An Admin chooses the Busy hours metrics under Show metrics. See Busy hours.
• When a human agent takes over a parked conversation, analytics count the pickup with its real wait and first response. Earlier days still show those conversations as unattended. See Team.
• AI reply times now run until a streamed reply is complete, so they can rise next to earlier days. See The AI replies drawer.

​

Agent builder

• A pasted image or video link in a message bubble shows as that image or video, never cropped. One save after an upload is enough. See Media in messages.

​

Tools

• A CodeKit action that uses a connected toolkit with deeply nested responses, such as Salesforce, now saves, and the editor types its nested fields. See Write a custom action.

​

Knowledge

• Deleting a knowledge base, or a retriever whose index no other retriever uses, removes the search index immediately. A restored retriever shows Needs indexing and answers again after a reindex. See Deleting is reversible and Restore a deleted knowledge base.

​

Live chat

• With Set auto offline on, a background dashboard tab that shows Conversations keeps you online and assignable. See Set auto offline.

​

Analytics

• For Admins, Members, and Auditors, the conversation drawer lists the conversations beside the full transcript of the one you select, with Transcript and Details tabs. See Drill into conversations.
• The Countries card lists a country once, by its two-letter code, for new conversations. See Countries and Devices.

​

Knowledge

• A website import skips a compressed page that unpacks to more than 10 MB, with the reason response too large, instead of failing the whole import. See Review crawl history.

​

Developer

• On paid plans, API keys from Settings → Integrations → API keys now start triggers and send webhook campaigns. Each key gets only the resources and permissions you choose. Keys generated in a Trigger gambit or a webhook campaign before this release no longer work, so create an API key instead. See API authentication.
• Webhook destinations send conversation, handoff, campaign, and end-user record events to your HTTPS endpoint, with retries, health status, and replay. See Receive events from Tars.
• The Contacts API replaces the end-user endpoints, with duplicate detection and batches of up to 50 records. The end-user endpoints are removed, so move those calls to the Contacts API. See Contacts API.
• Tars now also signs the requests that API Call gambits and webhook destinations send, so your server can check them. See Verify requests from Tars.

​

Security

• Admins manage the OpenRouter key in Settings → Integrations → Provider keys. See Provider keys.
• The identity verification key is on Signing secrets. Rotate key keeps the old key valid for 24 hours, and the card shows failed checks from the last 24 hours. See Verify end-user identity.
• Admins can copy a SIEM destination’s signing secret again from its edit dialog. See Stream audit events to your SIEM.

​

Agent builder

• {{system.transcript}} inserts the conversation so far into messages, Live Chat gambit messages, AI Agent instructions, API request bodies, custom code, and channel messages. See Insert the conversation transcript.
• A Trigger gambit lists Keys that can start this trigger, and Test trigger sends a test request with one of them. See Trigger gambit.
• API Call header values can be stored as secrets. See Secret headers and signing.

​

Campaigns

• A webhook campaign lists Keys that can send this campaign, and Create key opens the API keys page. See Webhook campaigns.

​

Widget

• The chat can pop out into a picture-in-picture window that stays visible when end users switch tabs. Turn on Pop out chat when visitors switch away on the Launcher tab. See Pop-out chat.
• AI agent replies start sooner after an end user sends a message. See Typing and streaming.
• When an AI agent writes text before a tool call, its answer after the tool starts a new paragraph in the same bubble. See Instructions.

​

Knowledge

• With Media search on, Test retriever shows ranked media on a Media tab next to the text results. See The Test retriever dialog.

​

WhatsApp

• Removing a 360dialog number now completes, and the number disconnects. See Open the number detail sheet.

​

Roles and access

• Admins, Members and Auditors can export data and resource configurations they can read. Audit log exports remain limited to Admins and Auditors. See Roles and permissions.
• Viewer and Billing roles see resource configurations and aggregate analytics without customer history or customer-data exports. Billing access still includes financial document downloads. See Roles and permissions.
• Members can read permitted conversations without a seat and manage conversations within their assigned teams and inboxes. Only Admins and Members can hold seats, which are required to reply to customers directly. See Live-chat overview and seats.
• Deleting a segment or a WhatsApp template needs the Admin role. See Roles and permissions.
• On Enterprise, Auditors can view SIEM destinations and delivery status. Adding or changing destinations, testing connections and replaying events remain Admin-only. See Stream audit events to your SIEM.

​

Analytics

• The Key metrics card replaces the Outcomes card. Admins choose its tiles with Customize, such as Open rate and Click rate for the launcher. CTR is now Click rate. See Key metrics.
• The Funnel steps are Visitors, Opened the widget, Started a conversation, Conversations, and Outcome. See Funnel.
• The Explore page is removed. See Analytics overview.
• Email reports show your Key metrics tiles and the new funnel steps. You can paste several recipients at once. See Email analytics reports.

​

Knowledge

• Media search follows each retriever’s media settings, including its result count and re-ranking. See The Build from scratch dialog.
• Experiment results show a Score for each image a retriever returns. See Measure retrieval quality.

​

Tools

• App tools and CodeKit now have a Logs view with tool call details, filters, and success rates. See Monitor tool calls.
• Tool calls show Live or Test, including test chats. The Component filter finds tool calls made by components such as the Meeting Scheduler. See Filter the calls.
• Test a workflow tool step with Test & map, then map response fields to variables. Delivery tracking is now Analytics purpose. See Enable and use actions.

​

Analytics

• The Tools & integrations section shows tool calls, failures, affected conversations, and connections that need attention. See Tools and integrations.
• Email reports now include Topics, CSAT, and a PDF copy when it can be generated. Reports use the chart palette saved by an Admin. See Email analytics reports.
• The Analytics page no longer has extra blank scrolling space below its content. See Analytics overview.

​

Deploy

• Each platform tab in Widget Embed has updated notes on where to paste the code. See Embed the widget.

​

Agent builder

• Card fields accept variables, and Auto populate creates one card per list item. See Card gambit.
• With Verify via OTP on, If verification fails picks the gambit where the flow continues after every verification code is used. The default is End without verification. See Text gambit.

​

Knowledge

• Each retriever has a 3 · Media retrieval section. Choose Inherit from text retriever or Custom media settings. See The Build from scratch dialog.
• Changed media is indexed automatically, including after scheduled crawls, and a retriever card shows Indexing media… until it finishes. See Configure a retriever.

​

Analytics

• Admins can schedule weekly or monthly reports in Settings → Reports, including recipients without Tars accounts. See Email analytics reports.
• Open a topic to see customer asks, answered share, AI handling, and findings backed by conversations. See Open a topic.

​

Agent builder

• Data Extraction fields now reach the agent, and their flow variables appear in the variable picker. See Data Extraction.
• Cmd/Ctrl+F uses browser find in text fields, the editor’s search in Code and API gambit editors, and the command palette elsewhere on the builder. See Builder basics.
• The builder header keeps the draft or live status and the gambit count beside the agent name again, and its controls stay centered. See Builder basics.

​

WhatsApp

• Taps on auto-populated menus and language menus route to the selected option, and conversations show full option labels. See End-user experience.

​

Agent builder

• Selecting a gambit in the palette shakes its Drag onto canvas to add hint, because gambits are added by dragging. See Builder basics.

​

Security

• Protected form answers remain usable as references by tools and component actions, including after conversation summarization. See Personal data protection.

​

Analytics

• Analytics cards clarify their counts, rates, comparison periods, and empty states. See Read the Analytics cards.

​

Security

• Person name detection protects Latin-script names in US and EU organizations. See Personal data protection.

​

Agent builder

• The builder’s Event Log shows test conversations again. See Test your agent.
• The AI agents page has a table or grid view, a Status filter, sorting, and an Open action. See The AI agents page.
• Rename, Duplicate, and JSON import and export are in the toolbar’s More actions menu. Validate sits at the top right of the builder. While you edit a draft, Preview draft opens it. See The toolbar.

​

Activity

• Large exports download in parts and retry a failed part automatically. See Export conversations.
• Component button labels wrap instead of being cut off. See End-user experience.
• Live chat status pills name the AI agent, for example Handed back to AI agent. See Human handover.

​

Live chat

• More actions in a conversation offers Export… to download it as CSV, JSON, or text. See Export a conversation.

​

End users

• An end user’s profile shows WhatsApp consent with Consent history. Change it with Opt in or Opt out, or for many end users with Set channel consent, and filter by WhatsApp consent. See Consent.
• The Users, Segments, WhatsApp numbers, and WhatsApp templates lists have Filter and Sort controls. See Profiles and Segments.

​

Knowledge

• Knowledge search returns faster. Each knowledge tool searches at most 10 knowledge bases. See Attach knowledge to an agent.
• A website crawl that imports no pages now says why, for example No pages matched the crawl filters. See Review crawl history.
• The Knowledge bases list has Documents and Index filters, sorting, and a table view. See The Knowledge bases page.
• A set of test questions whose images are being prepared shows Preparing media… with its progress and no longer stays stuck there. See Measure retrieval quality.

​

Agent builder

• Component gambits give agents native capabilities or run a component as a workflow step. See Component gambit.
• The Meeting Scheduler books through Google Calendar, Microsoft Outlook, or a CodeKit scheduling integration. Test chats simulate bookings. See Book meetings with the Meeting Scheduler.
• Generated files appear as download cards, and cards and buttons arrive faster and in order. See End-user experience.
• Web Search is unavailable and marked Coming soon. See Built-in AI capabilities.

​

Analytics

• The Components card reports component use, answers, failures, and meeting bookings. See Components.

​

Tools

• CodeKit actions have an Action type dropdown for general actions and integration roles. Toolkits and actions can be exported and imported as .codekit.json files. See Write a custom action.

​

Knowledge

• Free plans index with text-embedding-3-small. Reindexing with another model offers an upgrade or a switch to the supported model. See Configure a retriever.
• Datasets and experiments require a paid plan. See Measure retrieval quality.
• Existing website sources can re-sync when a free knowledge base is full. Existing pages update, and new pages need remaining capacity. See Import a website.

​

Activity

• Data exports include a Transcript column and conversation identifiers. See Export conversations.

​

Analytics

• Unknown sorts after named topics, and Manage topics shows the active topic count. See Find topics in conversations.

​

Analytics

• Classify again applies the current topics to conversations from the last 90 days, including Unknown results. See Classify again.
• Topic classification better matches the customer’s intent. See Find topics in conversations.

​

Widget

• End users can open the chat by selecting the launcher callout message, including with a keyboard. The close button dismisses only the callout. See Design the widget.
• Play a sound with the callout adds an attention chime, on by default unless you saved it off. It waits for page interaction and plays at most once every 24 hours. See Design the widget.
• Callout clicks count toward widget opens and CTR in Analytics. See Key metrics.

​

Knowledge

• A re-fetch of a website imported with hand-picked pages syncs only those pages. See Schedule re-fetches.
• Deleted website pages stay excluded from later crawls. Blocked pages lists them in Sync management, where Allow again makes a page eligible for import. See Import a website.

​

Live chat

• The inbox preview now updates when a long message is shortened at an emoji. See Work the inbox.

​

End users

• A WhatsApp message from a phone number that an existing profile already holds joins that profile. Tars no longer creates a second profile. This works for numbers saved from now on in international format, such as from a CSV import or the API. See End user profiles.

​

Analytics

• The new Topics card shows what your customers talk about, with each topic’s share of conversations. Admins add, merge, split, and archive topics in Manage topics. Topics are available on paid plans. See Find topics in conversations.

​

Knowledge

• Members can delete a website source from a knowledge base. Pages that only that website added leave Documents. See Import a website.
• A successful website crawl no longer reports Completed with errors. Failed and skipped pages show as counts, and the page count includes only the pages that were saved.

​

Agent builder

• Branches of one gambit that lead to the same gambit now show as one bundled edge. Select its label to disconnect one branch or all of them. See Builder basics.
• Fixed: View draft opens the draft chat again.

​

Security

• A protected response is no longer copied into the end user profile as a name, email address, or phone number. See Personal data protection.

​

Agent builder

• A Button option can create one button for each item in a list variable. See Button gambit.
• Auto Suggestion has a third source, From Variable, that builds suggestions from a list variable. See Auto Suggestion gambit.
• Button text and values accept variables. Tars fills them in when the agent sends the buttons.
• Fixed: picking a suggestion in an Auto Suggestion gambit no longer asks the same question again.

​

Security

• Choose how protected values look: mask characters, the rule name in brackets, or your own replacement text. See Personal data protection.

​

Analytics

• Analytics separates humans, AI agents that act for a person, and crawlers. AI agents are counted by default, and crawlers are not. See Analytics overview.
• The Trends card can compare rates for each published version of an agent. See Read the Analytics cards.

​

Knowledge

• A website source that starts at a section of a site, such as a docs folder, imports only that section. See Import a website.

​

Security

• Only the human agent assigned to a conversation can reveal a masked value, and only when an Admin turns on Allow reveal. Exports and transcripts keep the placeholders. See Personal data protection.

​

Analytics

• Analytics is one page of cards that an Admin can add, remove, and reorder for the whole organization. You choose which outcomes count, and the Funnel follows each widget impression to an outcome. See Read the Analytics cards.
• Goals and the Goals, Improve, and Team tabs are removed. Outcomes replace goals, and the Reply time, Handled by, and CSAT cards show how your team handles conversations.

​

Agent builder

• A Button gambit can hold up to 30 options. The builder shows 10 branch rows and collects the rest in a +N more chip. See Button gambit.

​

Campaigns

• A campaign can send each recipient their own values from a CSV that you import into a static segment. See Segments.

​

Live chat

• Fixed: a handoff at the start of a flow appears in the inbox right away. The notification sound plays on the end user’s first message.
• Fixed: the Mine tab no longer lists unassigned conversations.
• Fixed: private notes count as activity, so an active conversation is not closed for inactivity.

​

Agent builder

• Fixed: End gambits you add and trigger gambits can be deleted. The Start gambit and the default End gambit stay protected.

​

Deploy

• The widget header Name accepts up to 50 characters and the Subtitle up to 100. An oversized or unsupported logo or launcher icon shows an error. See Design the widget.

​

Chat experience

• Fixed: a pending verification code no longer carries into a new conversation, and a failed code delivery closes the code box.

​

Tools and CodeKit

• Fixed: a custom action can call two connections of the same toolkit, such as two Gmail accounts.

​

Security

• The widget reopens an earlier conversation only with proof of that session, such as its session token or a valid identity signature. See Identity signing.
• End users see AI reasoning and tool calls only when Show AI reasoning to visitors and Show tool calls to visitors are saved on for the agent. See What end users see.
• Fixes for security issues in the chat widget, file storage, and toolkit credentials. No action is needed.

​

Knowledge

• Fixed: files that finished uploading at the same time could stay in parsing or be marked as failed.
• Fixed: test question generation could stop early when several documents finished planning at once.

​

Live chat

• A message count in search shows + when more messages match than the dialog loaded.

​

Agent builder

• A conversation that an end user starts on WhatsApp fills the trigger variables with the opening message and the sender. A condition can now branch on trigger.userMessage. See Variables.
• The Store Value field of a No Input gambit accepts variables, such as {{enduser.phone}}. The value is resolved when the gambit runs. See No Input.
• Fixed: a gambit that asks nothing no longer stores a placeholder as the end user’s answer.

​

Reliability

• Fixed: in busy organizations, some agent replies and new conversations failed because of conflicting writes.

​

Knowledge

• Test questions for a large document are generated passage by passage and appear as each passage finishes. A large document no longer stalls the run.
• You can cancel a question generation while it runs.
• A test set holds up to 1,000 questions.

​

Knowledge

• Manage media shows images and videos as a grid of cards. You can hide an image or video so the agent does not show it in replies. See Manage documents.
• You can edit the surrounding page text that is embedded with an image or video, to remove noise.
• With media search on, the Test retriever dialog shows media results on their own tab. See Configure a retriever.
• An experiment cannot start while media indexing is still running, so every question is scored against the same media.
• Keyword and Hybrid retrievers rank results better. How rare a word is now counts within your knowledge base only.

​

Security

• Input gambits have a Protect Response switch. With it on, Tars protects the complete response with your personal data protection settings. See Personal data protection.
• In Mask mode, you can set original value retention in minutes, hours, or days, from 30 minutes to 365 days.

​

Activity

• The conversation feed shows the conversation from the end user’s side. End user messages appear on the right. See Watch conversations.

​

Knowledge

• An experiment scores each retriever on Recall, Precision, MRR, and Hit Rate. You set the weight of each metric.
• An experiment card shows the three best retrievers of the run.
• The question detail compares the expected answer with the retrieved text and marks the missing parts.
• When your default retriever already scores best, the result says that there is nothing to change.
• Run experiment stays disabled until the knowledge base has a retriever that is ready to answer.
• Fixed: a fully retrieved answer is marked found and gets correct metric scores. Run the experiment again to update an older result.
• Fixed: the best retriever of a run always shows in the ranking.

​

Agent builder

• Fixed: the help panel of the Custom Code gambit reads variables through variables, such as variables.response.Name. See Custom Code.

​

Data residency

• India joins the United States, the European Union, and Qatar as a data region. See Organization lifecycle.

​

Campaigns

• A campaign reply starts a new conversation even when one is already running, including while a human agent is in control. See Handle replies.
• The reply that started the conversation is kept in trigger.userMessage.

​

WhatsApp

• List and rating buttons read Show options and Choose a rating, and rating labels show on WhatsApp. See End user experience.
• A message longer than 4,096 characters is split into more than one message.
• Show knowledge citations now applies on WhatsApp.
• A message that could not be delivered shows in the Activity tab. See Watch conversations.

​

Security

• Form answers follow your personal data protection settings. See Personal data protection.

​

Agent builder

• API Call gambits send their headers with the request. See API Call.

​

Knowledge

• A new custom retriever gets a readable name from its search type and embedding model, such as Hybrid · OpenAI · 3-large. You can change the name. See Configure a retriever.

​

Tools and CodeKit

• Array and object inputs in Workflow Mode keep their structure when a referenced value contains quotes or line breaks.
• An input that is not valid JSON, or has the wrong type, stops the action before it runs.

​

Knowledge

• The Dataset tab is redesigned. Each test set shows as a card, and you can set document priority, question types, and dimensions when you generate test questions.
• You can edit a test question and select the exact answer text in its source document.
• Two test sets in one knowledge base cannot share a name.

​

Chat widget

• A video from a knowledge base plays directly in the agent reply.
• Card displays show images from a knowledge base.
• Captions on agent media replies show formatted text.

​

Tools and CodeKit

• CodeKit tool gambits run in Workflow Mode, with explicit inputs and success and failure routes. See Enable and use actions.
• A CodeKit action can be turned on only after it is deployed. The action picker shows the deploy state of each action.
• You can duplicate a CodeKit action or a whole CodeKit toolkit. A new toolkit starts with a Greeting action.
• The schema editor gains Form and JSON views and a Supported schema guide you can copy.
• Change account and Reconnect refresh the existing authorization first. A new authorization tab opens only when the provider rejects the refresh.
• A CodeKit action is retried only when the call failed before its code started, so a retry never runs the code twice.

​

Knowledge

• The Retriever tab shows shared indexes, build progress, and documents under repair.
• Retriever number fields show their allowed range.
• Validation flags a Knowledge Retrieval gambit whose retriever cannot answer yet, before you publish. See Attach knowledge to an agent.

​

Agent builder

• Input gambits gain an Allow Skip switch.
• Condition rules can compare the value of the gambit itself.

​

Chat widget

• A Redirect gambit that opens in a new tab shows a link for the end user to select.
• A resolved conversation leads to a new conversation in the widget.
• Widget Embed has a GTM tab with a Google Tag Manager snippet. See Embed the widget.

​

Live chat

• Active chat counts drop when a response timeout takes a conversation away from a human agent.

​

Data residency

• Qatar joins the United States and the European Union as a data region. See Organization lifecycle.

​

Knowledge

• You can edit a document title and body inline and group documents by source.
• A ready document no longer shows a status badge.

​

Agent builder

• Fixed: notes keep typed text, Ctrl+Shift shortcuts work, and Search Canvas opens again.

​

Security

• Tars support can erase an end user’s personal data on request. See Erasure and legal holds.

​

WhatsApp and campaigns

• Connect WhatsApp number offers two providers, Direct (Meta) and 360dialog. A number you already run on 360dialog connects with its API key. See Connect WhatsApp.
• The number detail sheet shows the provider of each number. For a 360dialog number, it checks where incoming messages go and can point delivery back to Tars.
• An Admin can copy a 360dialog number’s API key from the number detail sheet. Each copy is recorded in the audit log.
• Templates live in Campaigns → Templates and apply to the number you select. See WhatsApp templates.
• A template created from a campaign or the builder uses the number that sends it.
• A number import that fails points to a retry that runs the import again.

​

Security

• Admins can turn on personal data protection in Settings → Privacy. See Personal data protection.
• Protection finds card numbers, Social Security Numbers, email addresses, phone numbers, IBANs, and custom patterns in end user messages, captured fields, and CSAT feedback.
• Redact removes a value permanently. Mask stores a placeholder and keeps an encrypted original for 1 to 365 days.
• Card numbers and Social Security Numbers are always redacted.
• A member in live chat can reveal a masked value for 60 seconds, and each reveal is recorded in the audit log.

​

Agent builder

• Reasoning is off by default unless the model requires it. The Model parameters popover groups its settings and adds Max reasoning tokens.
• A parameter that the selected model does not support stays visible but disabled.

​

Knowledge

• When you upload files, you can pick, filter, and clear them. A file that is too large shows before the upload starts.
• A website import sets max pages and max depth up front. Advanced settings holds Include subdomains and User agent.

​

Billing

• The plan conversation allowance keeps renewing during a payment grace period.

​

Knowledge

• Knowledge bases open in a refreshed list and grid, with descriptions of up to 500 characters, document counts, and a create tile.

​

Agent builder

• The Language gambit gains an Allow Free Text switch.
• Turning Conditions off brings the option handles back.
• Fixed: editing a gambit no longer undoes connections drawn while its panel was open.

​

Fixes

• The conversation data card no longer lists city and country twice.

​

Chat widget

• Auto-read starts off on every visit and applies only to the current agent.

​

Knowledge

• A website import can send a custom User agent.
• Website imports capture content that loads late on JavaScript-rendered sites.

​

Fixes

• The signup phone field preselects your country.

​

Chat widget

• The toolbar on agent replies stays visible and shows the time. Read aloud and copy on the last message cover the whole run of messages.
• Multi-select cards collect picks as chips and send them together.
• Fixed: image galleries show public external image links again.

​

Knowledge

• With Media search on, an agent can find and show images and videos from a knowledge base. Manage media lets you add context to each item.
• Knowledge base videos play in the chat instead of opening as a link.

​

Agent builder

• Two new system variables, system.region and system.timezone, join system.country and system.city in the variable panel. The time zone uses the IANA format.
• Location variables are read when the widget page loads and stored when the conversation is created, so they are set from the first gambit.

​

Agent replies and in-chat input

• With File Reading on, an agent can answer from files the end user sends in the conversation. PDF, Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV, and plain text files are read. Images are described, with their text transcribed.
• A scanned PDF with no text layer is transcribed, so the agent can still read it. Video and audio files are not read.
• The agent reads a file by section instead of receiving the whole file, so a large document does not crowd out the conversation. Each read shows as a tool call when tool calls are visible.
• Files answered through a Media Upload gambit or a file upload component earlier in the flow are readable by a later AI Agent gambit in the same way.
• A file that cannot be read is reported to the agent with a plain reason, such as a password-protected or damaged file. A file over 10 MB is not read.

​

Chat widget

• Attached files upload as soon as they are added. Each file shows uploading, then its size, or failed with a retry message. Send waits until every upload has finished.
• A message carries at most 10 attachments of up to 10 MB each, and the composer names the file when one is too large.
• Fixed: the thinking indicator shows only while an AI Agent gambit takes its turn, and no longer on other gambits.

​

Agent builder

• A File Reading built-in tool joins the Built-in Tools & Capabilities dialog. It is off by default, like the other built-in tools. See Built-in AI capabilities.
• Fixed: Copy link on an agent card copies a working builder link. The old link opened an Agent not found page.

​

WhatsApp

• A document or image sent to a WhatsApp number is stored by Tars. An agent with File Reading on can answer from it like a widget upload.

​

Live chat

• Fixed: desktop notifications and sounds no longer fire for a conversation that is resolved or that the agent controls again. New assignments still alert you.
• Fixed: the agent response timeout no longer bounces a live chat back to a disconnected human agent every few minutes. It now hands the conversation only to a human agent with the dashboard open.
• When nobody is eligible, the conversation returns to the agent flow or waits unassigned. See Assignment and availability.

​

Security

• The dashboard and the chat host send a Content-Security-Policy header. Dashboard pages cannot be loaded inside an iframe on another site.
• The chat and widget pages still load inside an iframe on any site, so the widget script and the iframe embed keep working. The preview page can be framed only by the chat host itself and the dashboard, and every other chat host page refuses framing.

​

Agent builder

• Fixed: Custom Code gambits run again. Every custom code execution had failed with a sandbox error before the code started.

​

WhatsApp and campaigns

• A number alias holds up to 50 characters, a tag name up to 30, and a number up to 5 tags. Counters in the detail sheet show the current length.
• Length is counted in Unicode code points, so most accented letters and simple emoji count once, while a flag or skin-tone emoji counts more than once.
• Changing or disconnecting a number’s agent asks for confirmation only when an active or paused webhook campaign uses the number. The always-on re-route dialog in Distribute is gone.
• In the agent picker for a number, an agent already used by another number stays listed but disabled, with a Used by note naming that number.
• The number detail sheet is part of the page address. A reload reopens the same sheet, and closing it clears the address so back navigation does not reopen it.
• A template that Tars could not submit to Meta shows Template submission failed with the reason, plus Edit & resubmit and Dismiss failed attempt. A Meta rejection keeps its own Fix & resubmit action.
• A campaign whose header media is unavailable fails with media_missing and asks you to upload the media again before relaunching.
• Campaign delivery milestones are consistent. A delivery that reaches Read or Replied also counts as Delivered, so a reply never appears without a delivery.
• The webhook campaign API Contract card offers an importable Postman Collection and an OpenAPI 3.1.0 document next to the cURL example.
• A webhook campaign call with a recipient outside canonical international format fails with invalid_params and an errors entry of invalid_e164 on the to field.
• A webhook campaign send matches an existing end user by phone number with or without the leading +, instead of creating a duplicate end user.

​

Data residency

• The platform runs regional cells, and an EU cell now joins US. An organization’s data and its end-user traffic stay in that organization’s region.
• Widget and API traffic resolve the region from the embed. An organization that moves region keeps working with the embed code already on your site, with no change needed there.
• A missing or unknown region now fails with a clear error instead of quietly serving another region’s data.
• Knowledge retrieval keeps only vectors and identifiers in the search index. Document content is read from your own region’s database.
• Tars support places a new organization in the EU region and moves an existing one.

​

Agent builder

• The Conditions panel is available on the no-input gambit, with the same branch routing and Default fallback as other gambits.

​

Fixes

• A session that expires while the dashboard is open now returns to sign-in instead of an error screen.
• The error screen gains a Sign out button next to Try again.

​

Agent replies and in-chat input

• Agents can answer with interactive components instead of plain text. The set covers charts, data tables, cards, status lists, metric cards, progress trackers, timelines, accordions, code blocks, image galleries, and maps.
• Agents can collect answers inside the conversation using forms, button groups, confirmation dialogs, calendars, time slots, file uploads, and one-time code entry.
• An answered component freezes in place and keeps showing the answer, including on dashboard activity screens.
• Agent thinking is visible. A thought pill reports how long the agent reasoned, and each tool call shows its status and its payload.
• Agent replies gain a hover toolbar for read aloud, feedback, and copy. A link an end user sends unfurls into a preview card.
• Exit routes are editable. You can add, rename, describe, and remove a route, change where it points, and restore the default Complete route in one action.
• Paid plans can hide the Tars badge and set a footer disclaimer of up to 300 characters. Free organizations always show the badge.
• The escalate-to-human tool is retired, and human handoff now runs through exit routes. The standalone date picker is retired in favor of the calendar component.

​

Chat widget

• The widget opens straight into the chat. The home screen and the bottom tab bar are gone.
• The widget boots in a single round trip and preloads while the browser is idle, so it opens faster.
• A conversations list is reachable from the chat header. Rows are grouped by day, sorted by last activity, and can be swiped away.
• A returning end user can land on the conversations list instead of the chat. Turn on Open to conversations list in Distribute → Widget Embed.
• Spoken replies have a play control under each reply, and playback stops when the conversation changes.
• Fixed: typing indicators leaking between conversations, doubled question bubbles around components, and scroll losing the bottom of a conversation. A reply now quotes the message it answers, so the agent receives the right context.

​

Billing

• A Premium signup now grants the conversation volume it paid for. Some organizations kept the free allotment after upgrading.

​

Knowledge

• A knowledge base can now grade itself. The Dataset tab builds a test set from your documents, and the Experiments tab measures how well each retriever answers it.
• An experiment compares retrievers over one shared test set, so you can weigh recall against precision before you promote a retriever.

​

WhatsApp and campaigns

• Connecting or reconnecting a number shows live progress, keeps a background progress chip, and reports a failure you can retry.
• The Connect and Reconnect actions stay disabled until onboarding is available.
• Template submission is idempotent and shows Pending immediately. Header, body, and footer limits count Unicode code points, so accented text and emoji no longer miscount.
• Media rules are enforced on upload, replacement, submission, activation, and webhook send. Unhealthy and test-only accounts can no longer send a campaign.
• Campaign Activity reconstructs the conversation, pairing one outbound template message with the inbound reply.
• A webhook campaign dry run causes no delivery, no Meta call, and no credit movement.

​

Billing

• Invites are no longer blocked by a stuck seat cap.
• An invoice settled from a credit balance carries its real value instead of showing zero, and Enterprise credit grants complete correctly.

​

Organization and data lifecycle

• Data lifecycle now runs end to end, covering retention by plan, archival, organization-wide legal holds, and authorized archive exports. See Data retention.
• A legal hold blocks permanent deletion and archive expiry, while lossless archival continues.
• Test and preview conversations are permanently deleted after 30 days and are never archived.
• Deleting an organization runs as a bounded, resumable teardown that verifies no data remains before it completes.
• Onboarding no longer fails for a member who belongs to an organization but has not selected one.
• Idle conversations are closed by the abandonment sweep again, which had stopped running.

​

Knowledge

• Indexing no longer fails under load, and a stalled indexing job now recovers on its own.
• Indexing inside one knowledge base runs one document at a time. A single large knowledge base now indexes more slowly, and a faster strategy is planned.

​

Support access transparency

• Tars support access to your organization is now time-limited and visible. A banner shows who has access and the remaining time, and you can end access early.
• Support actions appear in your audit log as Tars Support, with the individual identity kept internally for accountability.

​

Knowledge lifecycle

• Retrievers now show a Needs reindex state with a Reindex all action, so index staleness is visible instead of silent.
• Imported documents can be edited in place, with revision history that records whether content came from an upload, an edit, or a website sync.
• Website re-fetch schedules warn when they cannot apply.

​

Fixes

• Fixed label creation on the conversations page.

​

Chat widget

• The widget’s file upload now enforces the configured maximum file size.

​

Conversations and delivery

• Agent replies, media, human-agent replies, and voice notes now deliver consistently on every active channel.
• Publishing reliability improved: live conversations always run the pinned published flow, never a draft in motion.

​

Campaigns and analytics

• Campaign screens were reworked for clarity, and delivery hardening reduces stuck or duplicated sends.
• Analytics gained drilldowns from summary numbers into the supporting conversations.

​

Tools

• Large toolkits load faster in the tool picker, and their action lists render reliably.
• Tool outputs can be projected to only the fields an agent needs, which keeps AI context small and answers focused.

​

Analytics

• Analytics gained metrics from page load through lead capture.

​

Knowledge and answers

• PDF image extraction is now plan-gated, and knowledge citation visibility is a per-agent design setting.

​

Billing

• Several Stripe purchase edge cases were fixed.

​

Analytics

• Daily analytics rollups now use consistent UTC dates.

​

Billing

• Plan pricing became plan-scoped, so organization-specific terms never leak between customers.

​

Onboarding and setup

• Organizations can upload a logo during onboarding.
• Sign-out is available from the onboarding screen, and invites to existing organizations work reliably.

​

Knowledge and answers

• Website import quality improved for hard-to-scrape sites.

​

Analytics

• Analytics was redesigned with a generic reporting system that covers campaigns alongside conversations.

​

Tars 3.0

• Organizations are now the home for every resource: agents, knowledge bases, tools, channels, campaigns, end users, and billing. Five roles (Admin, Member, Viewer, Billing, Auditor) and separate live-chat seats control access.
• The agent builder gains AI Agent gambits, exit routes, drafts and publishing, versioned rollback, and a test preview that is never billed. A rebuilt conversation engine runs every flow the same way across channels.
• Conversation variables move to ten clear namespaces with a pill-based insert menu. A test request can load real sample data, so field autocomplete works from actual responses.
• Tools become a first-class system: a directory of connected toolkits with OAuth and API-key connections, per-action enablement, and tool gambits with success and failure routes on the flow.
• CodeKit lets developers write, test, and deploy custom actions as organization tools. Custom code runs in an isolated runtime, and secrets stay separate from the code.
• Knowledge bases support website crawling with discovery preview, scheduled re-fetch, document editing, and multiple named retrievers with test-and-promote.
• The end-user experience was redesigned: streamed answers with clean partial rendering, generative interfaces such as forms, cards, charts, and date pickers, citations, and full keyboard and screen-reader support.
• Live chat is built in: a real-time inbox, assignment strategies, canned responses, CSAT, and a copilot for human agents. Human replies with rich content reach every active channel.
• WhatsApp connects through Meta Embedded Signup with template sync and campaign delivery from a prepaid campaign wallet.
• Analytics unifies agent, human, channel, and campaign results with drill-through to conversations and a build-your-own Explore view.
• Background work is unified: exports, imports, and bulk actions run in the background with visible progress in the task tracker.
• Security: a tamper-evident audit log with chain verification, SIEM streaming for Enterprise, bring-your-own-model-key, signed file downloads, plan-based data retention, and cold-storage archival.