What's new

Recent updates to Chatasist. This page is updated as features ship.

October 2026#

AI support agent#

• No more "sorry for the wait" on replies that weren't late (October 13). The agent used to open replies with an apology for keeping the customer waiting even when it answered within seconds, sometimes several times in one conversation. It now apologises for a delay only after the customer has waited more than five minutes, and only once. See Reply safety checks.
• One ticket per request: repeat requests are answered from the existing ticket (October 13). When a customer writes again about an order or parcel that already has a ticket — in a new conversation or from another channel — the AI agent no longer opens a second one. It tells them the request is in progress, or passes your team's reply on in the chat, and sends the ticket link again. A customer's follow-up goes onto the same ticket and a closed ticket is sent back to your team. Outside a channel's working hours, the agent also says when your team is next at work. Tickets now carry a reference code (the order or parcel number) that you can search and edit, and the customer's ticket page shows a ticket in a Done or Closed column as resolved. See Repeat requests.

Conversations#

• Filter conversations and tickets by date (October 14). Pick one day or a from–to range of days in the inbox, on a ticket board, or in the search across all your tickets — on its own or together with a search word. See Filtering by date.
• Search conversations and tickets by anything (October 14). The inbox search and the ticket search now find a conversation or ticket by chat ID, ticket number, customer name, phone number in any format, a customer code from the pre-chat form, or any word written in the conversation or ticket — an order number, "qaytarma", "gecikmə". The Tickets page has a search across all your boards. See Searching and Searching tickets.
• Conversations export is now an Excel report (October 14). The dashboard's export downloads an .xlsx file with each conversation's sentiment and ticket number, a second Daily statistics sheet with conversations opened per day by channel, and dates in your own timezone (09.10.2026 11:33) instead of UTC. See Export conversations.
• Reopen a resolved conversation (October 13). A resolved conversation used to be locked for good. It now has a Reopen button — where the message box would be, in the details panel and in the list's right-click menu — that brings it back as open with its history, assigned to you if nobody had it. Reopening does not undo the resolution: your resolved counts, AI-resolution usage and the rating all stay as they were. See Reopening a resolved conversation.

Tickets#

• Go from a ticket to its chat, and see a customer's tickets beside the chat (October 13). A ticket's details now show the ID of the conversation it was opened from, with a Go to chat link that opens it in the inbox in a new tab. In the inbox, the conversation's details panel has a new Tickets section listing the customer's tickets and where each one stands. See Ticket fields and Conversation details.

Channels#

• Customers can send pictures in the website widget from the first message (October 13). The image button is now there in a new conversation and while the AI agent is answering — before, a customer had to wait until someone on your team was assigned. Up to 5 images per message, JPEG, PNG, WebP or GIF, up to 10 MB each; anything else is left out with a short explanation in the customer's language. Files, videos and voice messages still become available once a team member is assigned. See Website widget.

August 2026#

AI support agent#

• Re-evaluated models, holding replies for slow lookups, and four reply safety checks (August 17). After testing candidates against real support conversations, the agent moved to models that reliably check your knowledge base before answering — nothing to configure, it's still managed AI infrastructure. When a question needs a live lookup from a connected MCP server (which can take 30–40 seconds), the agent now sends a short "checking" reply, the widget shows a Thinking… indicator, and the full answer lands in the same chat once the lookup returns — dropped cleanly if the chat is taken over or resolved first. And every reply is now run through four safety checks before it reaches the customer: it can't promise a follow-up it didn't schedule, can't read back personal details on an unverified chat, can't claim a false transfer to an operator, and has leaked internal text or looping repetition stripped. A fact-grounding check that compares numbers, dates, prices, and reference codes against what was actually retrieved is also available, off by default. See AI support agent and Resources & MCP.
• The agent answers instead of deflecting, and now knows who it's talking to (August 17). The AI agent's instructions were rewritten around how real support teams reply: it answers the actual question in its first message, keeps replies short and concrete, asks for one missing detail at a time, and offers a ticket only after a genuine attempt. It also collects a customer's email or phone naturally in conversation and — when those match an existing customer — confirms them with a 6-digit code sent to the address on file before opening that record. WhatsApp conversations skip the code because the number is already carrier-verified. See Customer identification.
• Simpler AI agent: answer or ticket (August 16). The AI agent now always does one of two things: it answers from your knowledge base, or — when it can't, or the customer asks for a person — it offers to open a support ticket. It no longer transfers chats to operators; your team can still view and take over any conversation from the inbox at any time. On a holiday, the agent mentions it's a non-working day and that ticket follow-up may be slower. As part of this simplification, the AI Health pages — the AI quality dashboard, knowledge gaps, operator coverage alerts, the weekly AI Health email, and "Train AI on this answer" — have been retired. See AI support agent.

June 2026#

AI support agent#

• New: holidays. Define non-working days — a single day or a multi-day span — under Settings → Holidays. On a holiday the AI agent stops offering to connect customers to a live operator: it explains it's a non-working day and offers to open a support ticket instead, while still answering questions from your knowledge base. The day is judged in your channel's timezone. Available on the Starter plan and above. See Holidays.
• Edit knowledge as a whole document, not chunk by chunk. Opening a file or pasted-text source now shows its full text in a single editor. Saving replaces the indexed text and re-indexes just that document in the background, rebuilding its chunks and overlap cleanly — so edits no longer drift between segments. For PDFs and Word files you're editing the extracted text, so you can fix extraction glitches in place. Available on every plan that includes a knowledge base. See Knowledge base.
• New: the AI quality dashboard. A new AI quality page (under AI in the sidebar) shows how your AI agent is performing over a 7-, 30-, or 90-day window — AI resolve rate, CSAT on AI-resolved chats, hand-offs to a human, instant cached answers, safety guard trips, and a breakdown of conversation outcomes. Available on the Growth plan and above.
• More reliable human handoff on Instagram, Messenger, WhatsApp & Telegram. On direct-message channels there are no tappable buttons beneath a message, so the AI agent now connects the customer to a human as soon as they clearly ask — instead of waiting for a button tap that may never appear. The website widget still presents the Connect / Create-ticket buttons first. See AI support agent.
• Bring your own tools to the agent via MCP. Need the agent to quote a price, look up a shipment, or pull data from your own systems? Connect an MCP server you host and the agent will call its tools during a conversation — only when relevant, and never fabricating a result if a call fails (Enterprise).
• Pick a record by tapping, not reading a list. When a customer asks about their data (for example their orders) and several records match, the widget now shows one tappable chip per record instead of printing a long text list. The customer taps the order they want to see its details, and can tap another to switch — a faster, less cluttered way to choose. See Resources & MCP (Enterprise).
• Record lookups now match your MCP tool automatically. When the agent looks up a single record (an order, a ticket, and so on), it reads your MCP tool's input schema and sends the identifier under whatever argument name the tool expects — order_id, ticket_id, ref, etc. — so lookups work without extra configuration. And if a tool reports an error, the agent now cleanly tells the customer it couldn't find the record rather than surfacing the raw error message. See Resources & MCP (Enterprise).
• The agent now uses the right tool for each follow-up question. If your MCP server exposes several per-record tools — say one for status, one for the item list, one for shipping — the agent picks the one that matches what the customer is asking. So after "what's the status of my order?" a customer can ask "what's in it?" or "where is it?" and get a real answer instead of a hand-off. See Resources & MCP (Enterprise).
• The AI now stays out of conversations your team started. The agent only joins a chat when the customer sent the first message. If you reach out first — replying to someone from the Instagram or Facebook app, or messaging a contact before they've written in — the AI no longer jumps in when the customer replies; the human who started the thread stays in control. See AI support agent.

Conversations#

• Sort your inbox, and flag conversations by priority. A new Sort control on the conversation list lets you order by Create time (the default), Last message, Unread first, or Priority. You can also give any conversation a priority of Low, Normal, High, or Urgent from the conversation header or by right-clicking it in the list — High and Urgent conversations show a coloured flag so they stand out. Available on every plan. See The inbox.
• Filter the inbox by assignee. The filter (funnel) menu now lets you narrow the conversation list to a specific teammate — or to just the Unassigned conversations — alongside the existing Status filter, which now lives in its own sub-menu. Each teammate is listed with their name and avatar. Available on every plan. See The inbox.

Channels#

• The website widget now speaks four languages. The visitor-facing chat widget interface — launcher, buttons, placeholders, tab labels, status and error messages — is now translated into Azerbaijani, English, Russian, and Turkish. Set the language with data-chatasist-lang on the embed script (az, en, ru, tr); any other value falls back to English, and the same value tells the AI agent which language to reply in. See Embed the widget and Localization.
• The AI agent's opening greeting now matches the widget's language. The first message the AI agent shows when a chat opens is now in the same language as the widget interface — driven by data-chatasist-lang, and defaulting to English when you don't set it — instead of your organization's first configured language. So a widget shown in English now opens with an English greeting. See Embed the widget.
• Chat messages now show relative times. In the open chat, each message's timestamp reads like "Just now", "2 min ago", or "yesterday" instead of a fixed date and time, and updates on its own as time passes. Times are shown in the widget's language. See Website widget.

Plans & billing#

• Updated monthly AI-resolution allowances. Included AI resolutions have been adjusted per plan: Starter 2,000, Growth 7,000, Scale 22,000, and Enterprise 70,000 per month (Free stays at 100). Need more on a given month? Top-up packs (+1,000 / +5,000 / +25,000) are still available. See Plans and Plan limits.
• Enterprise now runs on managed AI (no BYOK). The Enterprise plan no longer includes bring-your-own-key (BYOK); it runs entirely on the platform's managed AI models. BYOK remains available on Growth (optional) and Scale (default). See Plans and Plan limits.
• Cancellation flow is now a single, guided step. Clicking Cancel subscription in Settings → Billing opens a short flow that offers alternatives first — pause for a while, or switch to a cheaper plan — before you confirm. The cancellation itself still takes effect at the end of the current billing period; access continues until then. See Pause, cancel, or resume.
• Fairer AI-resolution counting. An AI resolution now counts only when the agent genuinely answered on its own. A chat that auto-closes after the AI couldn't help (a fallback or hand-off, then the customer goes quiet) no longer counts toward your allowance. And if a customer returns about the same unresolved issue within about 3 days, the conversation reopens and still counts as a single resolution instead of two. See Plan limits & enforcement.

Platform#

• Replies sent from the Instagram or Facebook app are now clearly marked. When a teammate answers a customer directly from the Instagram app or Meta Business Suite (instead of the Chatasist inbox), that reply is mirrored into the conversation with a Sent via Instagram / Sent via Facebook badge — so it's no longer mistaken for an AI Agent reply. See The inbox.
• Connect Instagram via Facebook (for threaded replies). Instagram now has two connection options under Settings → Channels: Instagram (sign in directly) and Instagram via Facebook (for an Instagram Professional account linked to a Facebook Page). The via-Facebook option delivers your replies to a specific message as native threaded replies in the customer's Instagram app. See Facebook & Instagram.
• See which login each Instagram channel uses. The channel list in Settings → Channels now labels every Instagram channel under its name — Instagram · Business Login or Instagram · Facebook Login — so you can tell at a glance how each one is connected. See Facebook & Instagram.
• Reply to a specific message. Hover any message in the inbox and click the new Reply arrow (just left of the ⋯ menu) to quote it. Your reply shows the original message stacked above it, and it's delivered as a native threaded reply on WhatsApp, Telegram, and Facebook Messenger — and quoted above your reply in the website widget — so the customer sees exactly what you're answering. (Instagram doesn't support threaded replies in direct messages, so there the reply arrives as a normal message.) Available on every plan. See The inbox.
• Continue with Google on the mobile app. You can now sign in to the iOS and Android apps with Continue with Google — the same one-tap sign-in as the web app. We create your account from your Google profile the first time, or sign you straight in if you've used Google before; if you already have a password account, you'll confirm that password once to link Google. Available on every plan. See Security & accounts.
• Stop the AI agent for a specific contact. Getting spammed by someone in the inbox? Open the conversation and use the new bot button in the header to stop the AI agent from auto-replying to that contact — across every conversation they start — while keeping the chat in your inbox for a human. Toggle it back on anytime. Available on the Starter plan and above. See The inbox.
• Reply from the mobile app. You can now reply to conversations directly from the iOS and Android apps — send text, attach photos or files, and record voice messages. Pull down on the inbox to refresh it. Available on every plan. See The inbox.
• Customers tab in the mobile app. The iOS and Android apps now have a Customers tab — an alphabetical, Contacts-style directory of your customers. Search by name, email, or phone, and tap a contact to see their details, channels, and conversation and ticket counts. Available on every plan. See Customers (CRM).
• Richer contact screen in the mobile app. The mobile contact detail screen now matches the web: one-tap Call, Email, and WhatsApp actions, at-a-glance stats (tickets, overdue, and average and total response time), a searchable list of the contact's tickets, the contact's conversations (tap one to open and reply), and a star for VIP contacts. Available on every plan. See Customers (CRM).
• Forgot password in the mobile app. The iOS and Android apps now have a working Forgot Password? link on the sign-in screen. Tap it, enter your email, and we'll email you a reset link — the same flow as the web app. See Security & accounts.
• Reset password link opens the mobile app. Reset emails now point at app.chatasist.com instead of the API domain. Open the link on a phone with the Chatasist app installed and it deep-links straight into the app on a "Set a new password" screen; open it on a desktop and it opens in the web app as before.
• Manage your account from the mobile app. The iOS and Android apps now have a redesigned Profile tab. Switch workspace, change language, edit your name and email, and update your password — each opens its own screen. You can also set up two-factor authentication: scan the QR code with your authenticator app and save your recovery codes, all from the app. See Security & accounts.
• Light, Dark, or System theme in the mobile app. The iOS and Android apps now have a working Theme setting under the Profile tab — pick Light, Dark, or System (which follows your device's appearance). Your choice is saved on the device. Available on every plan.
• Two-factor authentication now protects mobile sign-in. If your account has 2FA enabled, signing in on the iOS and Android apps now asks for your authenticator code (or a recovery code) after your password — the same protection the web app has. See Security & accounts.
• Status page — check the real-time health and 90-day uptime of every Chatasist service at status.chatasist.com. Review incident history and upcoming maintenance, and subscribe by email to be notified when an incident is opened or resolved. The page is public — bookmark and share it freely. See Service status.
• Status email notifications — you can now subscribe on the status page to get an email when an incident is opened or resolved. Sign-up uses double opt-in (confirm via a link first), and every email has a one-click unsubscribe. See Service status.

Channels#

• Turn voice or video calls off on the website widget. You can now hide the in-widget voice and video call buttons per channel from Settings → Channels → Website widget — offer voice but not video, or no calls at all on a given site. The toggles show on plans that include calls (voice on Growth and above, video on Scale and above), and your agents can still start a call from the inbox. See Turn voice or video calls off.
• Custom pre-chat form fields (Starter+) — add your own fields to the website widget's pre-chat form, beyond the built-in name and email — text, paragraph, email, phone, number, dropdown, or checkbox. Configure them under Settings → Channels → Website widget → Pre-chat form fields; visitors' answers are saved to their customer profile and shown to your team. See Website widget.
• Comment moderation & private replies (Scale+). On a forwarded Facebook/Instagram comment you can now Reply privately (sends a DM to the commenter), Hide/Unhide, or Delete the comment — straight from the message menu in the inbox. See Facebook & Instagram.
• Instagram story mentions land in the inbox. When a customer mentions your account in their story, it now appears as a conversation (with the story image) instead of being skipped.
• Message reactions (Facebook & Instagram). When a customer reacts to one of your messages (❤️ 👍 …), the reaction now appears on that message in the inbox in real time, and disappears if they remove it. Available on all plans.
• Replies sent from the Instagram app, Messenger, or Meta Business Suite now sync into Chatasist. If your team answers a Facebook or Instagram conversation outside Chatasist, that reply now appears in the inbox thread (marked as sent externally) and the conversation is treated as answered — so your history stays complete no matter where you reply. See Facebook & Instagram.
• Forward Facebook & Instagram post comments (Scale+) — comments on your posts can now be forwarded into the inbox as conversations, answered alongside DMs. Enable Forward post comments when editing a Facebook or Instagram channel, and optionally let the AI agent reply with a separate AI replies to comments toggle. Replies post publicly under the original comment, and comment threads stay separate from a customer's DMs. See Facebook & Instagram.
• Reconnect a Facebook or Instagram channel — the channel's ⋯ menu now has a Reconnect action that re-runs Meta authorization and refreshes the channel's permissions and webhooks. Use it so channels connected earlier can pick up new permissions (like comment management) without being deleted and re-added. See Facebook & Instagram.

AI support agent#

• Consistent handoff when no operator is online. When live handoff is unavailable (outside your work schedule or no operator currently online), the AI agent no longer offers to connect you to an operator and then show a ticket button. It now offers to open a support ticket directly, so the wording and the button always match. See AI support agent.

Admin & team#

• Invited teammates join instantly — no acceptance step. Everyone you invite now becomes an active member of your workspace immediately; membership is no longer gated on clicking the email. People who already have a Chatasist account (for example, a member of another workspace) can sign in with their existing credentials right away. Brand-new people are full members too — they just show as pending (meaning they can't sign in yet) until they set a password from the welcome email. The pending badge now reflects sign-in readiness rather than membership. See Team & roles.

Plans & billing#

• Clearer pause & resume. When you pause your subscription, the billing screen now shows a "Pauses on {date}" notice with a Resume subscription action, so the pending pause is obvious at a glance. Pausing is also safe to repeat — asking to pause again when a pause is already scheduled simply keeps the existing schedule instead of erroring. See Plans & pricing.
• Paused workspaces fall back to Free. Once a pause takes effect at the end of your billing period, the workspace is limited to the Free plan until you resume — paid features are suspended while billing is stopped, and Resume subscription restores your plan. See Plans & pricing.

May 2026#

Integrations#

• Automations and MCP servers moved to their own pages — Automations and MCP servers are now top-level items in the sidebar, just below Integrations, instead of tabs inside the Integrations page. The Integrations page now covers connections only (knowledge sources, CRM, action tools). Plans are unchanged: Automations needs Scale+, MCP servers needs Enterprise. See Automations and MCP servers.
• Search the Integrations page — a search box at the top of the Integrations page lets you filter integrations by name across every category (knowledge sources, CRM, action tools). Use it alongside the category tabs to find a tool faster. See Integrations.
• Notion knowledge sync (Growth+) — connect a Notion workspace and pick the pages the AI agent should answer from. Pages stay in sync automatically: edits in Notion re-index within minutes via webhook, and a daily safety-net job catches anything the webhook missed. See Notion.
• Confluence knowledge sync (Growth+) — connect an Atlassian Confluence site and pick the pages the AI agent should answer from. Install the Chatasist app from the Atlassian Marketplace for realtime updates, or rely on the daily sync. See Confluence.
• Confluence fix — page sync permission. Resolved an issue where some Confluence connections couldn't read page content and stopped syncing. Affected sites now show a clear prompt to reconnect, which restores sync. See Confluence troubleshooting.
• Faster knowledge sync. The background freshness check for all knowledge connectors (Notion, Confluence, GitHub, Google Drive, Zendesk, Dropbox, Slack, OneDrive) now runs about every 10 minutes instead of once a day, so content changes are picked up much sooner — even without realtime webhooks. See Integrations.
• GitHub & Google Drive knowledge sync (Growth+) — index READMEs/markdown from your GitHub repos and Google Docs/Drive files into a knowledge base. See GitHub and Google Drive.
• More knowledge sources (Growth+) — Zendesk Help Center articles, Dropbox files, Slack channel history, and OneDrive/SharePoint documents. See Zendesk Help Center, Dropbox, Slack, and OneDrive / SharePoint.
• CRM connector — HubSpot — let the AI agent look up customers in your CRM during a chat (Growth+) and create or update records from what the customer shares (Scale+). See CRM.
• More CRMs — Pipedrive, Salesforce, Intercom, Close, Apollo — the CRM connector now supports five more providers (sign-in for Pipedrive/Salesforce/Intercom, API key for Close/Apollo), same Growth+ read / Scale+ write. See CRM.
• CRM identity protection. The agent now only reveals stored CRM details — or updates an existing contact — for customers whose identity is verified (for example, signed in on your site through the chat widget). Anonymous visitors can no longer pull up or overwrite a contact by typing someone's email; at most a brand-new lead is captured. See CRM.
• Automations (Scale+) — define your own webhooks (Zapier, Make, or a custom endpoint) and let the AI agent trigger them mid-conversation — log a lead, open a ticket, notify a channel. See Automations.
• Action tools (Scale+) — first-class agent actions: post to Slack or Microsoft Teams, create Linear/Jira/ClickUp/Monday work items, look up Shopify orders, look up Stripe customers (with opt-in refunds), and check availability / share scheduling links via Google Calendar and Calendly. See Action tools.
• MCP servers (Enterprise) — connect your own Model Context Protocol servers and let the AI agent use their tools during a conversation. See MCP servers.

Plans & billing#

• AI-resolution top-ups — buy one-time packs of extra AI resolutions when you're running through your monthly allowance faster than expected, without committing to a higher plan. Top-ups sit on top of the bundled quota, are consumed only after it's used up, and stay valid for 365 days. See AI-resolution top-ups.
• Introduced subscription plans (Free, Starter, Growth, Scale, Enterprise) with a monthly AI-resolution allowance and region-aware pricing. See Plans & pricing and Plan limits.
• Annual billing with 2 months free.
• 7-day free trial on Starter, available once per workspace — try Starter before any charge, while higher plans bill from the start. See Free trial.
• Failed-payment recovery: a declined renewal is now retried automatically with email reminders to update your card, and your workspace keeps full access during the recovery window. See When a payment fails.

AI support agent#

• The AI agent answers conversations grounded on your knowledge base, with hand-off to humans, ticket creation, and CSAT follow-up.
• Configurable tone and per-channel enablement.
• The agent is now date-aware: it compares deadlines and validity periods in your content (e.g. a campaign "valid until 12 June") against today's date, so it won't present an expired offer as still active. See How grounding works.
• The agent now asks what the problem is before creating a support ticket, so every ticket records a real reason instead of opening empty. Connecting to a human operator stays immediate. See Tools the agent can call.
• AI Health rollout closes the feedback loop between what the AI couldn't answer and what your knowledge base needs:
  • New Knowledge gaps page clusters unanswered customer questions by topic, shows you which ones to fix first, and drafts a ready-to-edit FAQ entry that you can publish straight into a knowledge base.
  • New Train AI on this answer action in the chat menu turns a good operator reply into a knowledge base entry in two clicks. PII is scrubbed automatically.
  • New Operator coverage alert emails the workspace admin when too many handoffs to live operators went unanswered in 24 hours — so you can spot coverage gaps before more customers leave.
  • The weekly AI Health email now reports the AI resolve rate (with week-over-week delta), handoff and unanswered-handoff counts, and the top three active knowledge gaps.
  • On the knowledge base list, each row now surfaces Needs attention and Reindex due badges so stale or conflicting sources are obvious at a glance.
  • The Knowledge gaps page now supports filtering by Active / Resolved / Dismissed tabs, and drafted gaps link directly to the knowledge base the draft was saved into.
  • An in-app banner surfaces unread operator coverage alerts on every tenant page so admins notice them even when they don't check email.
  • Workspace settings gained a configurable operator coverage threshold (default 5 unanswered handoffs per 24h; set 0 to disable). Per-user opt-out toggle for the weekly AI Health email lives under profile notifications.
  • A daily LLM deduplication pass now merges near-duplicate gap clusters that the embedding-only detection left as separate rows.
• Agent-quality fixes (all plans):
  • The agent is blocked from claiming "I've connected you with an operator" when no one actually accepted the handoff — see No false transfer claims.
  • The conversation language is now pinned on the first substantive turn so later short messages or pasted foreign-script fragments don't make the AI flip languages.
  • When no operator is on shift, the Connect-with-operator chip is automatically replaced with Open-a-ticket so customers reach an async path.
  • PDF text extraction now tries multiple strategies and picks the best variant. PDFs that still look degraded (glyph drops, missing letters) get a Needs attention flag on their source so you know to re-export.

Channels#

• Connect WhatsApp (official Twilio API or personal number by QR), Telegram, Instagram, Facebook Messenger, the website widget, and email-to-ticket — all in one shared inbox.
• Hide "Powered by Chatasist" badge (Starter+) — new toggle in the website widget settings removes the footer branding. On the Free plan the badge is always shown.

Conversations#

• New workspaces now see a clearly labelled Demo conversation in the inbox until the first real message arrives — its sample messages now display correctly instead of appearing deleted.
• Fixed an isolation issue where users who belong to multiple workspaces could briefly see new conversations and notification-bell pings from another workspace appear in the workspace they were currently viewing. Realtime events are now strictly scoped to the active workspace.
• Fixed message ordering in the inbox: the "First contact" event could appear below the first customer message on Instagram, Facebook, Telegram, and WhatsApp when the webhook arrived a moment after the customer's send timestamp. Message timestamps now have millisecond precision and the event is anchored to the inbound message, so the order is always correct.

Admin#

• Delete workspace from Google accounts — administrators who signed up with Google can now delete their workspace by confirming with a fresh Google sign-in, instead of being blocked by a password they never set. This closes a gap where the subscription kept charging after trial because the workspace couldn't be removed. See Delete workspace.

Help center#

• Launched this documentation and help center, including a full API reference.

Browse the docs

Start from the introduction.