// 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 ads overview now lists every campaign once instead of duplicating Google campaigns across two cards, with diagnostics grouped in Account health. The conversion actions banner on the tracking goals page also stacks properly on narrow phone screens instead of squeezing into a word per line.
The AI output page now leads with a daily count of AI email replies, the share that carried the AI disclosure line, drafts waiting and social auto-replies, plus your ten newest AI replies linking to their conversations.
When requesting an AI edit on a creative, you can now ask for a voice-over to be generated alongside it, where this is switched on for your store. The editor tracks and applies the voice-over just like any other edit.
When you ask the AI editor to make a change, the chat now shows it thinking with a live timer and progress indicator, so you know your request is being worked on. Once it's done, you can see what version landed, and if it fails you can simply ask again.
When you add a video to a post, Magistry now generates a still image for networks that need one, like Pinterest, and shows it while it is being made. Photo-only networks such as LinkedIn and X can now use that same still as a photo post instead of being unavailable.
Competitor watches, tracking runs, and match link decisions can now be managed through the API and MCP tools, alongside a tracking budget that limits how many products and competitors a store can track. Actions that would raise your tracked spend or authority still require your approval before they take effect.
Your keyword waste, audiences, strategy, briefs, pixel health, suggestion queue, profit, marketing mix, and customer lifetime value data from the ads and performance pages can now be read through the API. If a platform isn't connected or hasn't synced yet, you'll get a clear message instead of misleading zeros.
The disputes settings, alerts, and identities pages now describe tracking, fraud, and notification behavior in plain merchant language instead of internal jargon. The alert network section replaces a non-working connect option with a request access link, since direct self-serve connections to networks like Ethoca and Verifi are not yet available.
The launch wizard's storefront stage can now show what Magistry set on your store, what it replaced, when, and why, where this is switched on for your store. The consent step also names your installed CMP or Shopify banner and how many clicks remain.
Help text, settings, performance, and run-error messages that used to reference LLM spend, model budgets, or USD now consistently describe your AI credit usage instead.
The conversation history panel in chat now matches the rest of the interface with a cleaner, unified look, better grouping of past conversations by date, and search that works as expected. Cards, chips, and surfaces throughout the chat also look much better in light mode.
The season plan now appears higher on the marketing page, with all twelve weeks shown at once instead of hidden behind a fold. Each week shows demand, planned budget, peaking categories and promotions, with the current week marked, where this is switched on for your store.
References to 'LLM' in help text, prompts, and panel labels have been replaced with clearer merchant-facing language across the ads wizard, performance panel, and customer service tools.
When a linked product's value was changed directly on a connected store, Magistry now holds the automatic sync instead of overwriting it, and queues an approval card explaining which store made the change. Approving the card applies the sync; choosing to keep the store's own value skips it.
Page content across the dashboard, including Settings and agent pages, now lines up consistently with the left-hand menu instead of drifting or sitting off-center on wider screens. Spacing is more consistent across banners, tabs, and connect pages.
The chat page now has a single consistent header, a readable thread with clearer message actions and tool-step summaries, and a composer that turns into a stop button while a reply is streaming. On phones, conversation history is reachable through a new drawer.
If you pause or enable a Google ad group or ad yourself, Magistry's automated writes no longer change that status, and you get one notice when it holds back.
The CS agent now withholds agent-written replies its detector reads as claiming to be human. Its money check was also re-tuned: fewer false blocks on vouchers and already-paid refunds, plus a check on parked amounts.
Actions like re-checking your store or fixing issues now show which step is running, how many steps remain, and how long it has taken, instead of leaving the screen unchanged or just showing a toast. Sourcing and catalogue searches also show a progress card and update automatically when the run finishes.
Paste a YouTube, Instagram or direct video-file link and confirm your rights, and Magistry fetches it as the remake source. TikTok links are not fetched yet.
You can now see and manage write permissions for connections to your store from Settings, including revoking access yourself. Each connection's allowed actions and spend are tracked for review.
After launch import, option names are translated into the store's language, imageless colour variants are linked to a matching photo where one can be identified, and single-colour mentions are trimmed from multi-colour titles.
The launch content check now counts every off-domain address on a page instead of the first two. Its message still names the first three problems.
If your store has opted in to the feed but has never had a push go out, you can now start that first push directly from the go-live panel instead of waiting for the next scheduled run.
When running the theme stage again, you can choose to redesign from scratch instead of reusing earlier design decisions, so a design that started off wrong is not carried forward. Theme design also now favours your live catalogue over a store profile that names categories you don't actually stock.
When a paid click from a Shopping or Performance Max campaign lands on a broken page, the health alert now correctly points to the Merchant Center listing instead of suggesting a final URL fix that would not apply.
The ads overview now leads with filters, key metrics, and the campaigns table, where it is switched on for your store. Campaigns no longer appear duplicated, and the table scrolls instead of being clipped on narrower screens.
The seasonal plan panel on the marketing page now shows a summary card with this week's demand, budget multiplier, next peak week, and peaking categories right away, instead of a closed, empty-looking fold. The full week-by-week table still stays tucked away until you choose to open it.
Feed destination cards now show the public link a destination registered and the platform's last fetch, where this is switched on for your store. An off switch alone no longer hides the Send button on API-registered destinations, only a genuine pause does.
Operator alerts now record an internal next-step route, such as reconnecting a data source or opening settings; it is not yet shown in the dashboard. Ad relevance also stopped calling a fix ready to apply when none is queued.
The tracking control on the tracking page is now a switch labelled with its current state, so it is clear what clicking it will do. Turning tracking off shows a message explaining that pixel events and purchase reporting to Google Ads will stop, and that message stays visible while tracking is off.
The briefs page can now show a plain-language explanation of why a live creative asset is not currently serving, where this is switched on for your store.
A new card shows how AI-driven visitors land on your product pages, split by holdout group, where it is switched on for your store. Before a measurement switch it shows a baseline period, then compares that baseline against days since the switch as a status through day 90.
You can now switch Brain's focus assortment, channel plan and launch playbook capabilities on or off on the plan page, and see who changed each one and when. Each plan area can also be set to Off: the Brain then proposes nothing in that area and runs nothing there, not even moves you approved earlier.
Magistry now records an 80% range on Brain actions where enough similar past predictions exist, and the plan page shows, per metric, how many measured outcomes landed inside it once there are enough to count. The range is not shown on each action yet.
You can now choose to have Magistry publish the new products it adds as drafts automatically, after a review window of the number of days you set. It is off until you set it, and deleting or archiving a draft in Shopify before then stops it from being published.
When Magistry remakes your video ads, text on signs and your product's artwork can now be tracked onto the footage from the real letters and your product photo's print, including lighting, folds and hands passing in front, instead of being redrawn by the AI edit. It is new, so check the text in a remake before you publish it.
When the person swap is switched on for a remake, Magistry replaces each person shown, not just one, and gives each speaker their own generated voice, and brand names spoken in the source are replaced too. Our team switches the person swap on per remake for now.
// 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
