Changelog

Every ship, in reverse chronological order.

New features, capability additions, integrations, and fixes — written by the team, surfaced here the moment we announce them in the dashboard.

August 2026
featureAug 30

Install steps built for the platform your site actually runs on

Getting the widget live used to end with the same line for everyone: paste this snippet at the end of your site's HTML. That is not a step you can follow on a site builder, where there is no HTML to paste it into — so the last five minutes of setup were the ones people got stuck in. Now, when we scan your site, we work out what it is built on — WordPress, Shopify, Wix, Squarespace, Webflow, Ghost, Framer, Duda or GoDaddy — and hand you that platform's real menu path instead. The instructions you email your developer are the same ones on your screen, so nobody is following two different sets. GoDaddy's Websites + Marketing gets its own snippet: that builder runs custom code inside a sandboxed frame, which trapped the chat bubble inside one section of the page instead of letting it float over all of it. And when we are not sure, we say so and let you pick — a confident wrong guess would send you down an install path that cannot work.

improvementAug 30

Deleting a persona now asks what to do with its knowledge

Deleting a persona used to quietly move all of its knowledge to Shared — which means every other persona on the brand could then quote it. Knowledge you imported for one prospect, pilot or demo could end up in answers to real customers, and nothing told you it had happened. Now the confirmation asks: delete the knowledge with the persona, or move it to Shared on purpose. There is no default — you choose. Large libraries are removed in the background, so the page never hangs waiting on thousands of items, and the dialog also spells out that a persona's email inbox history goes with it. The API takes the same choice via ?knowledge=delete.

featureAug 30

Visitors can fill in their details instead of being asked for them

When your brand requires details before booking a meeting or filing a ticket — a name, an email, a company — your AI no longer has to collect them one question at a time. In the website chat a small form now appears with a box for each detail still missing: the visitor types, taps once, and the conversation carries on. It shows up on the first reply when you’ve asked for details up front, and it rides along with a refusal, so “I need your email before I can file that” comes with the box to type it in. The form speaks the visitor’s language automatically — a French visitor sees E-mail and Entreprise, Arabic and Hebrew render right-to-left — with nothing to configure. Answers are saved exactly as if your AI had collected them: on the lead, on the conversation, notified to your team, synced to your CRM. Declining is always allowed, and on SMS, social and phone calls your personas keep asking the way they always have.

featureAug 24

Your personas can file support tickets — and your team works them in one inbox

Personas can now open a support ticket from a real conversation on any channel — web chat, social, SMS, or a phone call — and check its status when the customer asks. Your team works tickets in the new Tickets inbox: the full conversation alongside the ticket, per-brand ticket numbers, a status workflow, and confirmation and status emails to the visitor. Available as the Support Ticketing add-on — ask us to enable it for early access.

improvementAug 24

A shorter menu — six sections moved inside the pages they belong to

The dashboard sidebar is down from sixteen entries to eleven. Live chat, Social comments and Agent tasks are now tabs of your Inbox, so everything conversational lives in one place. Visitor memory sits under Leads as “Known visitors”. The Sales playbook joined Knowledge — which finally has its own spot in the sidebar — and Insights became a tab of Analytics, next to the numbers it explains. Nothing was removed: every old link and bookmark lands in the new spot automatically, and each moved page works exactly as it did. Fewer places to look, same product underneath.

featureAug 24

Quincer now reviews each persona's prompt against its real conversations

Once a month — or on demand from the persona page — Quincer measures each persona's instructions against its newest real conversations and proposes improvements you can audit: every signal cites the conversations it came from, suspected hallucinations are checked against the persona's actual knowledge, and suggestions apply as an additive line-level diff that can never delete your own instructions. Nothing changes unless you apply it.

July 2026
improvementJul 24

Facebook & Instagram DMs now show real names

No more 'Anonymous visitor' on social messages. When someone DMs your Facebook Page or Instagram account, we pull their profile name (and photo on Facebook) into the conversation, add new contacts to Leads automatically, and — if Visitor Memory is on — remember them across DMs and comments so your AI greets returning customers by name.

featureJul 24

Pilot, your in-dashboard assistant, learned 28 new tricks

The Quincer assistant in your dashboard can now read and act on far more of your account: browse your knowledge base, tasks, locations, phone numbers, voice transcripts, and suppressions — and, with your confirmation, edit knowledge items, take over (or hand back) live conversations, draft lead follow-up emails, update voice settings, and generate a sales playbook. It can also navigate you to the exact settings page or integration card you're asking about.

featureJul 24

Your AI can now walk visitors to the right page

Instead of just telling visitors where to click, your assistant can take them there. A visitor asking 'where is pricing?' can be navigated straight to the page — only ever on your own site (your domain only, so nobody can be sent anywhere unsafe). Your assistant also learns your site's deep links automatically from every crawl, including parameter pages. Available as the Page Navigation add-on — ask us to enable it.

featureJul 16

Your AI can now watch an Outlook inbox, not just Gmail

If your business runs on Microsoft 365 or Outlook, a persona can now monitor that mailbox and respond — exactly like it already could with Gmail. Point a persona at your Outlook inbox (sales@, support@, info@) and it reads each new message, searches its knowledge base and connected tools, then either sends a reply or saves a draft for your review — threaded correctly so the exchange stays tidy in the inbox. Connect it under Personas → Channels → Email inbox; monitoring is off by default on every persona until you turn it on. Gmail and Outlook now have identical inbox-monitoring capabilities. Scale plan.

featureJul 16

Bring your own MCP server

Plug any remote MCP (Model Context Protocol) server into Quincer — your own, or a vendor's hosted server like Jira or GitHub — and your AI personas instantly gain its tools, even for apps we don't integrate directly. Paste the server URL and an access token, pick which actions to enable, and they work right alongside your built-in integrations. Manage servers under Integrations → MCP servers.

featureJul 16

See what visitors ask for — and what you can't do yet

The new Insights page turns your finished conversations into a drillable knowledge map: every visitor request, grouped by type, sized by volume, and colored by whether your AI agent could say yes or had to decline. Zoom into a category, click a topic to read the actual asks, and see your Biggest Gaps — the requests customers most often hear "no" about — so you know exactly what to add next. Export the summary as a CSV or a shareable report. Scale plan; turn it on from the Insights page and it will analyze your last 90 days.

featureJul 13

Route each phone number to the right persona

You can now assign an inbound phone number to a specific persona, right from the persona's Channels tab. Calls to that number are answered by that persona — so a sales line reaches your Sales agent and a support line reaches Support, each with its own voice, hours, and escalation. The persona card shows the number(s) it owns at a glance, and any number you don't assign falls back to the widget's default persona. Add numbers under Integrations → Telephony, then assign them per persona (or provision them through the API).

featureJul 13

Choose which address each persona sends email as

A persona that watches a Gmail inbox can now send its replies from any verified send-as alias on that account — so an agent monitoring support@company.com can reply as sales@company.com, or each persona can send from its own branded address. Pick the alias under Personas → Channels → Email inbox (or leave it on the account default); only verified Gmail aliases are offered, and replies still arrive in the connected inbox. Microsoft 365 sends from the connected address, since alias-send there is controlled by your tenant admin. Scale plan.

featureJul 13

Your team's chats with a persona now live in their own Internal tab

When someone on your team chats with a persona in a Slack or Microsoft Teams DM, that's a colleague talking to your AI — not a customer. Those conversations no longer sit in Live chat waiting for a handover nobody needs, and they no longer clutter your Conversations list. They now appear under a new Internal tab, visible only to owners and admins; agents don't see them at all. Staff chats also stop triggering customer follow-ups, so a colleague asking a persona a question won't post a conversation summary to your history channel or email your team about it. They still count toward your plan's conversation usage, the same as any other conversation.

featureJul 11

No website? Quincer builds and hosts one for you

Small businesses without a website can now get one built by AI during onboarding — starting with real estate agents. Answer a short interview about your business, and Quincer designs a complete single-page site, hosts it on your own address, and embeds your AI assistant already trained on your answers. Edit anything in the new Website section: change copy, upload photos, or just tell the AI how to redesign it (with full undo history). Publishing is part of the Website + Widget plan; building and previewing your draft is free.

featureJul 11

Talk to your AI persona right inside Slack

Your team can now chat with a persona directly in a Slack channel — not just receive escalations. Bind a persona to a channel and any message there, whether a plain post or an @mention of the bot, is answered in-thread by that persona using its knowledge base and connected tools, posted under the persona's own name and avatar. It works like an internal help desk: ask "what's our refund policy?" or "draft a reply to this lead" right where your team already works. Set it up under Personas → Channels → Slack → Assistant channel (invite the bot to the channel first, and reconnect Slack once so it can post as the persona). Every answer counts as a conversation, exactly like a chat on your website, WhatsApp, or SMS.

featureJul 11

Give each persona a face — avatars for your AI agents

Every persona can now have its own avatar instead of a coloured initial. Pick from 20 built-in avatars, upload a real photo, or generate one with AI — or keep the initial-on-colour default. The avatar follows the persona everywhere: on the Personas screen, next to every conversation so you can see at a glance which agent handled it, and live in the chat widget itself (the header and each of that agent's message bubbles). Set it per persona under Personas, or via the API. A small touch that makes multi-persona setups far easier to read.

June 2026
improvementJun 26

Pick a persona for each social channel

You can now choose which persona handles each connected Facebook Page, Instagram account, and WhatsApp number — right from Integrations → the connection's Account tab. Run a playful, on-brand persona on Instagram and a buttoned-up one on Facebook, each with its own system prompt and tools. The assigned persona answers that channel's comments and DMs; leave it on "Widget default" and nothing changes. Find it under Integrations once a Meta channel is connected.

featureJun 26

Sentiment escalation: auto-flag conversations that turn negative

Your assistant now watches conversation sentiment and escalates the ones going sideways. When a visitor sends several negative messages in a row, the conversation is automatically routed to a human — flagged on the Live Conversations dashboard, or posted to your Slack, Teams, or Telegram escalation channel with a Take over button. Set the negative-streak threshold and a cooldown per persona under Persona → Channels → Sentiment escalation. Requires sentiment scoring (Analytics → Sentiment). Scale plan.

featureJun 26

Monthly Performance Report: your AI's ROI, in your inbox

On the 1st of each month, your account now gets an executive performance report by email — conversations handled, resolution rate, median first response, CX score, a 6-month trajectory, and an estimated support cost saved with every assumption shown. The ROI is deliberately conservative (it only counts resolved chats that would have reached a human), so it holds up to a finance review. It also surfaces the top content gaps worth closing and recommended next moves. Tune the cost assumptions and preview last month's report under Settings → Reports.

featureJun 16

Privacy controls: automatic data retention & deletion

You can now control how long Quincer keeps personal data and respond to data requests, all under Settings → Privacy. Set an automatic deletion window per data type — visitor memory, chat & voice transcripts, voicemail recordings, leads, surveys, and inbox email — and Quincer deletes anything past that window daily (files included). Need to handle a data-subject request? Look up, export, or permanently erase everything held about one person — by email, phone, or visitor ID — across every store. Every deletion is logged for your compliance records.

featureJun 16

See your customer satisfaction at a glance

Your Analytics → Feedback tab now gives the full CSAT picture: average rating, resolution rate, and — new — your response rate (how many conversations actually left feedback) plus a responses-over-time trend and a per-persona breakdown. Averages built from only a handful of ratings are flagged as directional so a couple of responses never mislead you. Turn on the CSAT survey from that tab to start collecting it.

featureJun 16

Know how every conversation feels

Turn on Sentiment scoring (Analytics → Sentiment) and Quincer classifies every visitor message — positive, negative, a question, or neutral — and rolls it up per conversation. The new Sentiment tab shows your message-sentiment mix, average sentiment over time, and a "Needs attention" list of the conversations with the longest run of negative messages, so you can jump on a frustrated visitor. It runs on the lightweight utility model, off the chat's critical path, and is off by default.

May 2026
featureMay 31

Visitor & team memory: your assistant now remembers people

With Subject Memory on, your assistant recognizes returning signed-in visitors — greeting them by name, recalling what they asked about last time, and skipping questions they've already answered. It also remembers social commenters, and gives the dashboard Pilot recall of your team's recent changes. Off by default, opt-in per widget, verified-identity only, with per-person export and delete (GDPR).

featureMay 31

Custom Tools: connect any REST API to your assistant

Give your assistant new abilities by connecting your own REST APIs as tools — no code on our side. Part of a redesigned Integrations dashboard that makes connecting Slack, HubSpot, Calendly, Google, and more far clearer.

featureMay 31

Microsoft Teams: bring your AI assistant into your channels

Connect Quincer AI to Microsoft Teams so your assistant can answer questions and hand off to humans right inside your team's channels and 1:1 chats — with per-persona channel routing and a dedicated history channel.

featureMay 1

Manage leads from inside Perplexity Computer

Ask Perplexity Computer to show your hot leads, take over a live chat, or update lead stage — in plain English. Actions are attributed to a dedicated Perplexity team-member identity in your Quincer inbox, so you get a full audit trail instead of anonymous bot activity.

featureMay 1

OpenClaw integration: one-click connect, full API

Your OpenClaw agent can now drive Quincer end-to-end. One-click Connect OpenClaw in the integrations page spins up a dedicated OpenClaw team member and issues a scoped API token — no webhook setup, no manual role assignment. The token powers a new /api/v1 surface covering conversation takeover, reply, and handback plus full leads and activities read/write. A downloadable SKILL.md drops into ~/.openclaw/workspace/skills/quincer/ and OpenClaw is live.

featureMay 1

Your voice widget now speaks 30+ languages

Visitors can talk to your voice agent in their preferred language and get responses back in the same language — no cascading speech-to-text / translation / text-to-speech pipeline, no extra latency. Configure supported languages, set a default, and turn on auto-detect in Widget → Customize → AI agent. English, Spanish, French, German, Japanese, Arabic, Hindi, Mandarin, and more.

Want every ship in your inbox?

Subscribe to the monthly update and we'll round up everything that landed.

Start freeRead the blog