v1.9
Bigger uploads and better personas
- Improved Question history opens to the chat behind every question. Each kind of question now has an Open to see the questions button, and every question in the list ends with Click to go to chat, which opens the conversation it came from.
- New Synthia goes full screen. The ⤢ button in Synthia's header opens the assistant across your whole screen, with the conversation in a comfortable reading column. Press it again, or Esc, to go back to the small window. Synthia remembers your choice on the next page.
- Improved Question history in every language. The Question history tab now reads in all 13 dashboard languages: the menu, the browser tab, every kind of question and the figures beside them. Your customers' own messages are shown exactly as they wrote them.
- Improved LINE WORKS hand-overs say exactly why the bot did not answer. When the bot passes a chat to your team because the workspace is out of credits or its billing is paused, the hand-over says so and tells your team what to change. Your customer still sees the same short message.
- New Topics written from your own chats, in your language. AI reads your whole chat history and names up to 15 topics in the words your customers use, in every language they write in, and every message is counted against them. Japanese questions now have their own topics too. Sentiment insights and topic summaries are written in the language your dashboard is in.
- Improved Topics and Sentiment now read your whole chat history. Topics counts every message your customers have sent your agent, not only the most recent ones. Sentiment reads conversations from across your whole history and adds four charts: sentiment over time, sentiment by channel, customer ratings and conversation length.
- New Your data, on your terms. In Workspace settings you can export everything as one file, delete your account, and choose how many days conversations are kept. On the Deploy page, Reissue link replaces a link you no longer trust. Admins can keep every conversation with the agent's own AI provider.
- New Mask personal details, per workspace. Turn it on in Workspace settings and the email addresses, phone numbers and card numbers your visitors type are replaced with [redacted] before they are saved or reach the AI.
- Improved Security hardening, reviewed against ISO/IEC 27001, ISO/IEC 27017 and NIST CSF. A bot looks up billing, orders or tickets only for a visitor who has signed in, and only their own. An agent that requires sign-in protects its pictures and documents too. Two-factor secrets are encrypted at rest, the activity log is tamper-evident, and every release runs a dependency and credential check.
- Improved Off-site backups run once a night. The daily snapshot now runs at midnight Japan time instead of throughout the day, so it never competes with your traffic, and it can be encrypted with AES-256-GCM.
- New One main switch, then choose where your AI support works. Under Share your chatbot as a link, turn the main switch on, then choose Link and SharePoint, Link only, or SharePoint only. With SharePoint only, the link does not open but SharePoint keeps working. The SharePoint section shows which is on, and its check button tells you whether the agent can be seen there.
- New SharePoint has its own switch, separate from the share link. Until now the SharePoint embed was the share link inside a frame, so switching the link off also switched SharePoint off. Channels now has a SharePoint switch: turn it on and the bot keeps answering inside SharePoint while the share link stays off. You can also limit it to your own SharePoint site. Turning the link off now tells you whether SharePoint stays on.
- New Your LINE bot can answer from your AI agent's knowledge base. The menu item is now called AI Agent LINE Bot. On a bot's new Knowledge base tab, choose one of your AI agents. Messages to the bot on LINE, and on its other channels, are then answered from that agent's sources, with its instructions, safety rules and credits, and each conversation also appears in the agent's inbox. The tab lists the agent's sources so you can see what the bot knows. A bot with no agent linked keeps its built-in answers.
- New Synthia speaks with ChatGPT voice, and suggests questions about your own fleet. Voice to voice with Synthia now uses ChatGPT voice in every language, including on the Synthia page, which has its own 🎙️ button. Replies read aloud and the chat widget voice also use ChatGPT voices by default; an agent with its own ElevenLabs voice keeps it. Both can be changed in Admin → Settings → Voice. Synthia’s suggested questions are now built from your own agents, such as where customers are unhappy or what is being handed to a person.
- New When the assistant hands you to a person, it now tells you what actually happens next.
- Improved Every Matrix AI product is now in the footer, in your language. The footer lists all sixteen Matrix AI sites, from Call and Studio to SalesLens, Hawk AI and MEIO, plus the full product list. Where a site has a version in your language, the link opens that version.
- Improved One language number across the whole site. Different pages quoted different figures — some said “30+”, the guide said “Japanese and 13 others”, and the homepage counter said 14 — which left you doing arithmetic to work out what we actually support. Every page now says 14: the languages your assistant and the dashboard are supported in. Nothing about your agent has changed, and it still accepts and replies in whatever language a customer writes in; 14 is the number we stand behind and support.
- Improved Every channel webhook was re-audited, and secret comparisons are now uniform. Each of the forty inbound addresses your channels deliver to — Meta, WhatsApp, LINE, Slack, Telegram, Teams, Twilio and the rest — was sent unsigned and wrongly-signed traffic. None of them acted on it: no message arrived, no credit was spent, nothing was switched on. Three places compared a secret using an ordinary equality check, which can reveal a little about the secret from how long the comparison takes; all three now use a single constant-time comparison, which also refuses to match when a channel has no secret configured yet. Card payment confirmations remain switched off until the signing secret is in place, rather than being trusted unverified. Nothing about connecting a channel changes for you.
- New A setup checklist for your first agent, and a short review before it answers customers. The dashboard now shows six steps — add knowledge, read five answers, set the limits, give it one thing it can do, connect a channel, go live — with a progress bar. Each step opens the page that already owns that setting; nothing is set twice, and a step counts as done only when it really is, so deleting your last source puts that step back. Progress is stored with your workspace rather than in one browser, so it follows you between devices, and “I’ll finish later” hides it without losing anything — a one-line Resume brings it back. The one step that is more than a tick: before a new agent can be connected to a channel, somebody has to read five of its answers in the playground and confirm they are answers you would be happy for a customer to receive. It takes about a minute, and it is the only thing standing between a wrong answer and your customers. On a paid plan an owner can skip it by accepting the risk, which is recorded with their name. Agents that are already connected to a channel or have had conversations are unaffected — this applies to agents that have never been live.
- Improved An agent on a model your plan no longer includes now answers on the standard model. If your plan changes and an agent is still set to an advanced model, it keeps answering your customers on the standard model from the same provider, charged as the standard model. Your setting is not changed: upgrade and the agent goes back to the model you chose.
- Improved Your agent now finds the right note more reliably. A question that names a note, such as “refund policy”, now goes straight to that note instead of a longer page that mentions the same words. For website visitors, a note that is nothing but meeting links and passcodes is left out entirely, and links in note titles are removed too. The playground shows what the knowledge base returned for each answer.
- New Risky messages are now flagged for review in Chat logs. Every incoming message on every channel is checked for shared keys or passwords, card numbers, lists of personal data, attempts to override the assistant, and common scam patterns. A match puts a ⚠️ on the conversation in Chat logs with the reason. Nothing is blocked and the reply still goes out; only the reason is recorded, never the message.
- New Summarize a conversation in one click. The 📝 button in Chat logs writes a short summary of the conversation: what the customer wanted, what they were told, and anything still open. It uses only that conversation, never adds facts, and removes meeting links and keys before anything is sent to the AI.
- New Choose Synthia's voice yourself, including OpenAI speech-to-speech. Open the ⚙ in the Synthia panel to pick how she listens and speaks: the standard voice, OpenAI speech-to-speech, or OpenAI listening with the natural voice speaking. You can also choose the OpenAI model and voice. Your choice is saved to your account; Site default keeps following the settings chosen for the site.
- Improved Every key on Settings → Login & keys now says whether it actually works, not just whether it is filled in.
- Improved Every key on Settings → Login & keys now says whether it actually works, not just whether it is filled in.
- New A reply answered from a web search now uses double credits. When your agent searches the web to answer a question, that reply is charged at twice the usual credits, and so is each message of it if it is split. Replies from your knowledge base cost the same as before. The multiple can be changed in Admin → Settings.
- New Web search can now run on Claude or ChatGPT, with no separate search key. When a LINE WORKS question is not in your knowledge base, the web search can now use Claude's or ChatGPT's own search, on the AI keys you already have. Leave the provider blank and it tries Claude, then ChatGPT, then a Serper, Brave or Google key if you added one; if one fails, the next one answers. Before this, a key saved with the provider left blank searched nothing.
- New LINE WORKS: answers from the web when your knowledge base has nothing, and from the makers' own pages for product questions.
- New A data-deletion page, and automatic removal when you remove the app on Facebook. The new Data deletion page explains how to remove what you connected with Facebook, Instagram or WhatsApp, and how to delete your account. If you remove Matrix AI Chat from your Facebook settings, Facebook now tells us and we delete the access tokens you granted and switch that channel off automatically.
- Improved Synthia and the chat widget work better on phones.
- Improved Synthia now knows your workspace's whole conversation history, and speaks your language everywhere.
- Improved When someone adds you to their workspace, you now see it straight away.
- Improved The how-to guide now shows the real screen for far more steps.
- New Campaigns can now be sent on LINE, LINE WORKS, WhatsApp, Telegram and Slack.
- Improved The how-to guide is up to date with everything shipped since mid-September, in all thirteen languages. 45 new walkthroughs cover several workspaces and invitations, the chat log at scale, asking Synthia about your chat logs and the product, the whole LINE WORKS Behaviour panel, and the personal-data memorandum when you create an agent. 69 existing ones were corrected — most notably the sidebar item that is now called Channels, and channel setup that starts there — and 20 that described controls the product does not have were removed. The LINE WORKS Behaviour panel itself is now translated, and a walkthrough Synthia suggests on her own page now runs on screen instead of showing a code.
- New You can now own several workspaces and move between them from the header. Open the workspace menu (▾) next to the workspace name and choose New workspace — for a second business, or a sandbox kept apart from production. Each workspace has its own agents, team and plan, and the same menu switches between every workspace you own or were invited to. Menu items with a badge beside them, such as Outbound, are now translated like the rest of the navigation.
- New Synthia now knows the whole product, and answers how-to, feature and plan questions from its own documentation. Every answer she gives can draw on the site’s pages, the support knowledge base, all the how-to walkthroughs and the changelog — in your language where the product has been translated — and she names the page it came from or runs the matching walkthrough on your screen. When the documentation does not cover something she says so rather than guessing. New features are included automatically as they ship.
- New When you are invited to another workspace, the app now tells you and switches you there in one click.
- New Synthia now reads your chat logs and tells you what to do with them.
- New LINE WORKS behaviour now matches the customer’s Worker v41, and always shows which version is running. New in Channels → LINE WORKS → Behaviour: requests only a person can do (a workflow change, a shared mailbox, an account) skip the assistant; 「できない」 in a first question is answered, and only a short 「できない」 right after an answer escalates; support hours in Japan time with public holidays computed; one support room per person, reused instead of opening another each time; trigger rules with an exclusion that beats everything; a test box and a step-by-step self-test. The panel always shows the engine version, the server build and when the settings were saved, 「バージョン」 in the talk returns the bot’s status, and replies can end with the build that answered. Pasting the Worker keeps the handler ids you already saved.
- Improved The chat log now shows up to 10,000 conversations at once, and the conversation you open stays on screen while you scroll.
- New Settings → System now shows the two background jobs that decide how fast your agent answers and how large your backups are.
- Improved Every agent with real visitors now gets the head start, not just the first five. We asked the service how many agents it actually has conversations for, rather than guessing again, and the answer was nine — nine agents on the whole service have spoken to somebody other than their own owner in the past week, while the limit was five. Four were being left out for no reason, and the last one that did make it had been idle three and a half days, so this was never a crowded queue. The limit is now twelve, which covers every eligible agent today with room to grow, and the service status reports the eligible count so the day it passes twelve is visible rather than discovered by someone waiting. The important part is not the number: with the limit above the number of agents that qualify, which one is considered “most recent” stops mattering at all, and that ordering is exactly what the two previous attempts got wrong.
- Improved The first person to use your agent after an update no longer waits three times as long as everyone else.
- New Slack installs with one approval, and stays off until it could actually answer. Connecting Slack meant creating your own Slack app, installing it, copying a bot token out of one page and a signing secret out of another, then setting an events address by hand — four steps, each of which this site can do once on everyone’s behalf. Where the site has its own Slack app, Slack now connects the way Facebook does: press Add to Slack, approve it in your own workspace, and your agent starts answering mentions and direct messages with nothing to copy. It asks for four permissions and no more — exactly the ones it acts on — so nobody is granting access to channel history that nothing reads. There is one thing it deliberately refuses to do. Slack signs every message it sends us, and the key for that signature belongs to the app rather than to your installation. Without it the install still succeeds: a real bot appears in the workspace, the card reads Connected, and then every message Slack delivers is discarded unread. A bot that is connected and deaf is worse than one that is plainly not connected, because nothing about it looks wrong. So the button does not appear until that key is in place, and where it is missing the page says which one it is instead of offering an install that would have ended that way. Connecting by hand still works throughout, for anyone who has their own Slack app and would rather use it.
- New WhatsApp connects through Meta’s own sign-up, and says plainly when it cannot.
- New Prospects whose address we could not read are now reachable.
- New Every LINE WORKS inquiry now gets a reference number. N-HNK plus six digits, shown at the end of the reply, at the front of the room name the bot opens, and in the handover post. One inquiry keeps one number; after a handover, or two hours, the next question is a new inquiry. A restart can skip a number but never reuse one.
- New The chat log now shows who you are talking to.
- New The memorandum is now readable in all thirteen languages, clause by clause.
- New Invitations are written in the language of the business you found.
- Improved Sending mailboxes are set up for Namecheap Private Email out of the box, each with its own password. A mailbox’s SMTP settings now arrive filled in for Namecheap Private Email — the outgoing server mail.privateemail.com on port 465, which you can change if you host elsewhere. Because every mailbox on the same provider still signs in with its own password, you can give each one its password on its own line when adding a batch: put the address, a space, then the password, one mailbox per line. Each account is then stored with its own credential and joins the rotation sending through its own server. A line that is only an address, or a comma-separated list of addresses, still works and sends as a From header through the shared provider, and a single password typed into the SMTP field still applies to any line that has none of its own.
- New Add a whole list of sending mailboxes at once. The Sending mailboxes panel in Outreach took one address at a time, which is a long afternoon when you are warming up a pool of a hundred accounts. Paste the whole list now — one address per line, or separated by commas — and every valid one joins the rotation in a single step, sharing the sender name and daily limit you set alongside. They start taking their turn immediately: each invitation goes out from whichever account is furthest from its own daily limit. An address already in the pool is updated rather than added twice, a malformed line is reported and skipped instead of stopping the rest, and adding a single address still works exactly as before.
- Improved Japanese walkthroughs animate over the Japanese dashboard again. Each step of a walkthrough is played over a real screenshot of the screen it describes, with the control it names ringed — and when the ring would land outside the picture, the player correctly falls back to a drawing rather than pointing at the wrong thing. A third of the Japanese measurements were in that state: they had been taken with the sidebar’s groups expanded while the screenshot beside them was taken with those groups collapsed, so 248 steps silently reverted to the drawing and nothing reported it. The Japanese screens have been re-captured — 1,382 steps now show the real product, up from 458, and not one rectangle falls outside its own picture. A check now fails if a future capture produces any.
- Improved Every walkthrough now reads in your language.
- Improved Japanese is now spoken from its reading, not from its characters. A kanji does not carry its own pronunciation — 生年月日 can be read せいねんがっぴ or, character by character, as something no one says — so the voice was guessing, and the wrong guesses are the first thing a Japanese listener notices. Every Japanese line is now converted to kana before it reaches the voice, because kana is the pronunciation itself: 設定 goes out as せってい, 既読 as きどく, 代理店 as だいりてん. Counters and dates, which were consistently wrong, follow the sounds the language actually makes: 1件 as いっけん rather than いちけん, 10分 as じゅっぷん, 1月1日 as いちがつついたち, 1人 as ひとり. Nothing on screen changes — you still read 設定 — and the readings come from this product’s own Japanese, so the words it says most often are the ones it says best. Three deliberate limits: katakana and Latin are left alone, since katakana is already a pronunciation; a lone kanji whose reading depends on the words around it is left for the voice to read in context rather than guessed at here; and nothing that is not Japanese is touched, because Chinese is written in the same characters and reading it as Japanese would produce confident nonsense.
- New Agree to the personal-data memorandum before creating an agent.
- Improved The signing requirement is enforced by the server, not only by the page that shows it.
- Improved Synthia now knows how to read Japanese aloud, not just which language to speak. Being asked to speak Japanese is not the same as being told how to say things. Counters are now read correctly (3件 as さんけん, not “three ken”), dates use Japanese readings (1月1日 as いちがつついたち), English product names are said the way a Japanese speaker in that field says them rather than forced into katakana, and the register stays consistent instead of drifting between polite and plain forms. Chinese received the same treatment for numbers and measure words.
- New Talk over Synthia and she stops.
- New Hear an OpenAI voice before you choose it.
- Improved Synthia’s own knowledge base is now the same screen your customers get for theirs.
- New Pick Synthia’s voice from a list, and hear it before you save. Her voice was set by typing a name into a text box — “Claire - Ultra Real & Natural” — and the languages that should sound different were set by writing lines like “ja = …” into another. A name that matched nothing on the account fell back to the previous voice silently, so a typo could only be found by hearing the wrong voice. Settings › Voice now shows every voice on the account, filtered by language or by country, with a play button beside it so you can listen before you save, and sliders for speed, stability and similarity. Beneath them is one row per language carrying both the engine and the voice, because those two decide the answer together: put a row on an engine that speaks through OpenAI and it offers OpenAI’s voices, since the ElevenLabs voice on that row would never have been used. And choosing an engine that needs speech-to-speech while speech-to-speech is switched off now says so on the page, with a button to turn it on, instead of leaving it to be discovered at the microphone.
- New Give a customer more team seats without moving them to another plan.
- Improved The chat widget now checks its answer before it shows it to you.
- New Search and filter your agents.
- New Twenty-one more, and now sixty-three business systems. Opsgenie and Datadog so “is it slow for everyone” is checked rather than guessed; Kayako and TOPdesk for ticket status; Asana for work that is not a code change; Attio, Copper and Zoho CRM so it knows who it is talking to; Greenhouse and Lever to tell a candidate where their application stands; Personio, KING OF TIME, SmartHR and Garoon for an agent your own staff use; Adobe Commerce, Squarespace and Wix for shops; Klaviyo and Brevo for “am I still on this list”; PayPal disputes, so a buyer is answered before a dispute becomes a chargeback; and Metabase, which runs a question your team already saved rather than writing SQL at all — and only that one question, because a bot that could choose its own report could read every report you have.
- New Fifteen more, and now forty-two in total. ServiceNow for enterprise incidents; Chargebee and Recurly for subscriptions; BigCommerce and Square for shops; Trello, ClickUp, monday.com, Redmine and Backlog for “where has this got to”; a Notion database queried as a database rather than read as prose; Outlook Calendar and Acuity for what is actually free; Mailchimp for “am I still on this list”; and BambooHR for an agent your own staff use, which answers from the directory and nothing else — no pay, no absence, no personal address.
- Improved What the agent can WRITE is now one honest line instead of fifteen implied ones.
- New Sixteen more systems your agent can look things up in.
- Improved A calendar lookup never reads an appointment’s title out loud.
- New Your agent can sit in a Telegram group without answering everything in it. Add the bot to a group and it stays quiet until it is addressed: an @mention, a reply to something it said, or a /command aimed at it. Everything else in the room is left alone, because a bot that answers every line is not a support agent — it is the reason the bot gets removed. Its answer is threaded onto the question it answers, so in a busy room nobody has to guess what it replied to. Two cases it gets right that are easy to miss: a /command aimed at a different bot by name is left to that bot, and a reply to a different bot’s message is not a reply to ours. Telegram’s own privacy setting already withholds most of that traffic, but it is a setting in your BotFather account that anyone can turn off, so it is not what this rests on.
- Improved Four things on the “building next” list now say what is actually in the way. WhatsApp groups, Discord forum channels, Matrix and Reddit were listed with a description of what they would do and nothing about why they were not done. Three of them are not waiting on us in the ordinary sense: the WhatsApp Cloud API publishes no group messaging at all, so that one needs Meta rather than time; Discord forum posts and Matrix both arrive over a persistent connection rather than a webhook, which is a different shape of thing from every channel we run; and Reddit publishes no webhooks, so it needs a poller. Each row now says so. “Coming soon” about something nobody can build is a promise, and it is not one we can keep.
- New Five more places your documents already live. Google Drive, OneDrive, Dropbox, Box and an Obsidian vault join SharePoint and Notion as things the agent can be trained on and re-read as they change. Google Docs and Sheets are exported rather than downloaded, because they have no file to download — without that, a Drive full of Docs indexes as nothing at all while reporting a successful sync. The folder walk is breadth-first on purpose: read depth-first, it spends the whole budget inside whichever folder sorts first, so the handbook at the top level is missed because an archive folder came before it alphabetically.
- Improved A sync now tells you what it did not read. Every one of these has limits — how many files, how much text, how deep — and a sync that quietly stops at one of them produces an agent answering confidently from a fraction of your handbook. It now names the files it skipped and the ones that were too large, so a folder that is half PDFs and half spreadsheets tells you which half the agent learned. An empty result is reported as a failure rather than a success, and says which formats it can read. A folder id that does not exist says so, instead of reading as a permissions problem and sending you to the wrong settings page. Obsidian is the one worth a sentence of its own: it publishes no API to read a vault through, so this reads the Git repository the vault is kept in — which is what people actually do — and the fields say so rather than leaving you looking for a setting that is not there.
- New Your agent can use the integrations you connected, instead of being told about them.
- New Your own API can be one of those lookups.
- Improved The connections diagram answers you.
- New Answer in the help desk you already run, and from the wiki you already write.
- New Eight more places your agent can answer, and nothing to build on your side.
- Improved A provider that has run out of money is no longer asked again every thirty seconds. When an AI provider fails we stand it down briefly and use the next one, so a passing blip does not cost every reply a wasted round trip. Thirty seconds was right for a blip and wrong for the failure we actually hit: OpenAI answering “you have no credits remaining”, which nothing but a payment can change. Every customer message went on paying for that call — the wait before it gave up, and the providers that do work pushed further down the queue behind it. An account that is out of money, or a key the provider rejects outright, is now left alone for ten minutes instead, because both need a person rather than a wait. Neither is permanent: topping up, or entering a new key, takes effect on the very next message without a restart, and a genuine traffic spike still clears in thirty seconds as before.
- Improved Every integration we have not built yet now shows its mark. The channels you can connect today were each shown with their logo. The list underneath — the one with the “I need this” buttons, whose entire job is to tell us what to build next — showed the first letter of the name instead. A hundred and forty-eight grey initials is a wall rather than a list, and the eye slides straight off the part of the page we most want you to read. Each entry now carries a mark: the brand’s own where one is published for this purpose, and otherwise a drawn glyph of what that integration would actually do. Nobody’s logo is approximated, because a near-miss of a familiar mark reads as carelessness about the thing itself.
- New Send outreach from several mailboxes, and never past what any one of them allows.
- New Synthia can be taught from files, pasted text and web pages, not just a text box. The facts she should always know were a single field, which is right for three standing facts and useless for a manual. She now has a knowledge base that works exactly like an agent's data sources: upload a PDF or Word document, paste a long note, or give her a page to read. She looks these up when a question touches them rather than carrying them in every conversation, so there is no practical length limit. The text box stays and still works — short standing facts belong there, stated word for word; documents belong in the knowledge base. Nothing you add is published: it has no share link and no visitor can reach it.
- New You can insist on the assistant, even on a subject that normally goes to a person.
- New Some subjects go straight to a person, and the assistant is never asked.
- Improved The escalation phrase is now a list, and it understands how people actually type.
- New Build a demo from a link you paste, without searching for the company first. The Prospector finds businesses by searching. Often you already know the one you want — you are looking at its Instagram, or its entry on Google Maps — and searching for something you can already see is busywork. There is a “from a link” tab now, and each kind of link is handled by the only method that can honestly work for it. A website is read the way a searched-up company is read. An Instagram or Facebook profile gives its name, its bio and its picture, which is all a profile publishes to anyone who is not logged in. A Google Maps place is looked up properly through the same search the worldwide tab uses, so the phone number and address are the real ones rather than guesses scraped from the link. A search results page is refused outright, and says to open the business and copy that link instead — building from one would have produced a demo for Google.
- Improved A generated demo now opens in the company's own colour. Every demo came out purple, which is ours and not theirs, and the one thing a demo has to do in its first second is look like it belongs to the company it was built for. Where a site states its colour — the one browsers paint the address bar with — the assistant is created in the nearest of our eight themes. Where a site states nothing, or states white, it stays purple on purpose: most sites carry a white address-bar colour because a tutorial told them to, and reading that as a brand choice would pick a colour by rounding error, which looks like a mistake rather than a placeholder. The colour is only ever taken from what a site declares about itself, never guessed from the pixels of its logo.
- New Text a business their own demo, the moment it is built.
- New Forty-three more kinds of business you can prospect for.
- Improved A worldwide search can now return up to a thousand businesses. The old ceiling was two hundred, which is short of a single category across a whole city. The ceiling was also written down twice — so raising the one you could find changed nothing at all. What a large number really gets you now sits next to the control, honestly: Google Places returns at most sixty results per search and Foursquare fifty, however many are asked for, so going beyond that means having OpenStreetMap switched on. The time budget for the map search grows with the size of the request too; before, a large sweep hit an internal thirty-second limit and came back looking as though the service were down.
- New The staff list on a sign-in-protected link downloads as a real spreadsheet.
- New Importing a staff list now asks what it should do to the list you already have.
- New Choose who gets the sign-in email, and what language it arrives in.
- Improved And the person helping you is told when that happens. A time limit that only you can see would mean us still editing while you are editing too — the same collision from the other side. When the half hour is up their screen says so, says you have your dashboard back, and stops there. Coming back in is a deliberate act, and starts a fresh thirty minutes.
- New Reach a business that published no email address.
- New Synthia speaks a different voice in each language. The browser told us which language it was reading in on every spoken reply, and that was thrown away — so one voice read all fourteen, and a voice chosen because it sounds right in English also read the Japanese. You can now set a voice per language, and write it the way you read it in the picker (“Claire - Ultra Real & Natural”) rather than as a 20-character code. Anything without its own entry uses the site-wide Synthia voice, as before.
- Improved Settings → Voice shows which voice each language is really using.
- New Every AI provider now says whether it is working, and tells you before the credit runs out. A key was only ever shown as “set” — which a revoked key, a key from the wrong account and a key with no credit left all look like, and all fail in front of a customer. Each provider now carries its live status beside it, and the dashboard shows a warning as soon as one stops working, without anyone pressing anything. The check runs hourly on its own, because the failure worth catching is a card expiring at 3am, and until now nobody found out until they happened to open the console.
- New An email when a provider fails or runs low.
- Improved Files up to 100MB.
- Improved Two large uploads at once no longer fight each other. Reading a big PDF is the most expensive thing the service does, and several at the same moment could exhaust the memory of the whole machine. They now queue: two are read at a time and the rest wait their turn, which is slower for the third file and safer for everybody who is not uploading anything.
- New A list of everyone waiting for a feature that is still being built. Pressing “Notify me” sent one alert at that moment and then stored the request out of sight, which is no use on the day the feature ships — by then the alert is a thousand messages up and nobody can answer “who did we promise this to?”. There is now a Waitlist tab listing every request with the address we would actually write to, and a button that copies them. It shows two things a plain list hides: the people whose account has no email address at all, whom we can never answer, and the people whose address has changed since they asked — writing to the address stored with the request looks exactly like keeping the promise while reaching nobody.
- New A warning before your trial ends, not after.
- New A weekly summary of what your agents actually did. Conversations, messages, how many agents were live and which day was busiest, once a week. A quiet week sends nothing at all — a digest reporting zero conversations is just a weekly reminder to cancel.
- New Emails in Malay and Romanian. Both are languages the site has always been readable in, but an email addressed to someone who reads them arrived in the base language instead. They are now offered everywhere a language is chosen for a member, an invitation or a shared document.
- New Select agents and act on all of them at once.