// changelog
Every entry is written when the change ships, by the people who shipped it, and it is the same list the product shows inside your dashboard.
// the record
The Brain, the ads specialist, the theme engine, the storefront agent and the customer-service desk each carried their own mode, and they disagreed on live stores — one desk was set to autopilot in one place and to dry run in another. There is one control now; each agent's page says what it inherits and links to the setting that changes it.
You can now add your own logo to the support portal your customers use to track and reply to tickets. Upload an image in the portal branding settings and it will appear in the masthead straight away. A preview of how the portal looks to customers is available on the same page.
A new control on the ticket page lets customers flag their request as urgent. On stores with a staffed inbox the thread is held for a person; on stores handled automatically the request is prioritised rather than paused, so it still gets a reply. The button's confirmation survives a page refresh, so the flag is never silently lost.
When a ticket is closed, customers now see a satisfaction rating panel at the bottom of the support page they were already linked to — no second email required. The rating appears only after a ticket is closed, and a submitted score is shown in place of the form on return visits. The follow-up rating email is no longer sent when the portal page can collect the rating instead.
When Magistry reads your policies and proposes knowledge rules, any rule that needs information from you — like a return address or store-specific detail — now shows an editor so you can fill it in before the rule goes live. Rules are grouped by how much attention they need, conflicts with your published policy are called out separately, and a search bar reads your own wording, not just our paraphrase. Approving in bulk no longer pre-accepts rules that flagged a genuine disagreement with your store's policy.
When your support team has put a refund or compensation offer on the table, you can now see the amount and its deadline directly on your ticket page — no need to dig through your inbox. Two reply shortcuts let you respond in your own words, and your reply is what confirms your decision.
When a customer writes in about an order, agents now see a panel with the relevant order details — including the carrier handover, delivery status, and any refund settlement — pulled directly from what was already sent to that customer. The panel only appears when the order can be matched with confidence, so agents never see a guess. Product photos from the order are shown inline.
When a customer opens their support ticket in the portal, any photos or files attached to the conversation are now visible to them — and they can attach a file of their own when replying. Previously, a filename was shown beside a message but the file itself could not be viewed.
Cleaning up outdated offers from your feed now runs with higher concurrency instead of one at a time, so large backlogs that previously took over an hour of wall-clock time finish in a fraction of that. Quota limits are still fully respected, and the count of removed versus not-found offers is reported accurately as before.
When a customer opens their support ticket link, they now see who owes the next reply and how long ago that happened — rather than a raw message count. The message body shows readable text by default, with the original email available on demand. The page also now discloses when replies may be drafted by an automated assistant, as required.
You can now connect your Reddit Ads account and manage campaigns alongside your other ad channels. Conversion events are tracked through Reddit's Conversions API, and the AI assistant can plan and dispatch Reddit ad actions just as it does for other platforms.
New stores now receive a personal offer to book a one-on-one onboarding call directly inside the welcome email. A named onboarding specialist reaches out so you can get set up quickly and with support.
You can now reach the Magistry team by email — and book a call when a scheduling link is available — without leaving the dashboard. The contact card appears on Settings → General and the Help & guides page.
When Magistry checks your support playbook against your published pages, it now runs a second verification pass before marking anything as a conflict. Conflicts that were really just your playbook choosing one valid option from several your site already offers are quietly downgraded instead of blocking your rules. Real contradictions — like a page showing a phone number while your playbook says phone support is unavailable — are still caught and flagged.
Filing applied the folder's tag and left the conversation in the inbox, so a filed thread sat in two places. Tags label and folders file, and the two are now different things. Routing rules can file automatically, they never overrule a folder a person chose, and every rule application is recorded with its outcome.
New stores now receive a dedicated customer service portal hostname automatically, without anyone needing to run a manual setup step. The hourly job mints, registers, and verifies the host, and the portal link appears in the dashboard once the host is confirmed live.
Every outbound reply email now includes a personalised link so the customer can view their ticket and conversation history directly, in their own language. The link is only included once the store's support portal has been verified as reachable, so customers will never land on a broken page.
When your playbook is compared against your published store pages, claims are now checked in smaller groups so contradictions that were previously missed now surface correctly. The panel also no longer claims to catch every disagreement — it tells you what it actually does, so you know to treat the results as a strong signal rather than a complete audit.
You can now upload your own support playbook — a Word file, PDF, or web page — directly from the CS Knowledge page or the CS setup wizard. Magistry reads your document, sorts each section into the right part of your agent's training, and flags anything that conflicts with your published store policies for you to review before it goes live.
A dedicated customer portal lets shoppers open their support ticket by following a link in any email you send them, read the full conversation, and post a reply — all without logging in. Each store's portal lives on its own permanent address that won't break if you ever rename your store. Email bodies are displayed safely, with no risk of one store's content appearing under another's address.
Your customer opens one page on your own subdomain that holds the whole conversation, the photographs they sent, a reply box and a rating. It is branded per store, it works without an account, and a ticket with nothing in it is refused rather than shown as an empty conversation.
The AI assistant you use across the dashboard, your merchant emails, and the mobile app is now called Magistry everywhere you see it. No features have changed — only the name on the screen.
The AI that drafts customer service replies was learning from automated form letters, customers' quoted text, and emails mislabelled as agent-written — none of which reflected how your desk actually writes. It now studies only the messages your team composed, with quoted conversations and HTML template wrappers removed before any comparison. Draft tone and phrasing should more closely match the voice your customers already recognise.
When the brain's conflict guard is holding a campaign slot, you can now see exactly which move is blocking, how long it has been held, when it will release at the latest, and how many proposals have been turned away during that window. Landing-page quality signals are now grouped by destination URL, so the brain can tell the difference between pages that need work and pages that are already clean.
When a cost spreadsheet can't be matched to your catalogue, the import assistant now explains exactly which identifier and cost columns it tried, what it was able to read from your catalogue, and which other columns in your sheet could work instead — so you have a real next step rather than a dead end. If the file genuinely has no other way to name a product, it tells you that plainly too.
// direction
// stay current
// changelog list
// e-mails not sent yet · no tracking pixels
// see it run
Connect your store and choose the Brain's mode — in dry run every decision is logged with what it would have done, and nothing touches your store.
The Brain changes nothing until you choose a mode · Every action logged · Price, status and copy changes can be rolled back
We use cookies to run this site, to see which pages help, and to measure which ads bring people here. You choose per category; everything except the strictly necessary is off until you say otherwise. Cookie policy · Privacy
