Homepage Chat lands as the new front door
For Anyone who opens Glowtify with a half-formed idea — operators, marketing managers, agency users describing a brief for a client store — who want to start in conversation rather than picking a template and filling a form.
A chat lives on /home as the primary way to start a campaign. The expanded chat input is open by default (no click-to-expand), supports multi-channel campaign briefs in a single thread, and now pulls structured context from Connectors and a section-based Business DNA route plus multi-CTA support. Pasting a content reference triggers an AI extraction pass on the URL so the model has more to work with, and a Back link returns you to the greeting + feed view without losing your conversation history.
Why it matters: Starting a new campaign meant deciding upfront where you wanted to publish, switching surfaces to brief the AI, and re-pasting context each time. There was no single place to walk in to and say "here's what I want to talk about" in plain language.
The Home feed grows up — richer cards, calmer updates, real team collaboration
For Anyone who lives in /home — marketing managers and account managers tracking a queue of campaigns across channels, plus reviewers and assignees who want one place to see everything waiting on them.
Cards on /home now show the channel's brand icon (Meta, LinkedIn, Klaviyo, etc.), the task's first image as a thumbnail, and for email tasks (Mailchimp / Klaviyo / CyberImpact) a rendered email preview. SCHEDULED cards carry a live countdown ("in 4 hours") that flips to overdue when past, and the CTA relabels from Reschedule to View task once the time passes. New items no longer silently shuffle into your view while you're reading — a centered "N new item(s) — refresh" banner appears and you decide when to load them. Every item gets a one-click Star (with a Starred tab badge counting how many you're tracking), and items are only marked read when you actually engage with them. On the collaboration side, when a teammate submits a task or epic for approval, mentions you in an epic comment, or assigns you to an epic, the corresponding APPROVAL / MENTION / ASSIGNMENT card now lands in your home feed alongside the existing email — with a Reply CTA that deep-links straight to the right panel.
Why it matters: Last week the home feed shipped, but it was easy to miss what was scheduled vs published at a glance, you couldn't bookmark anything to come back to, and the feed quietly rearranged itself while you were reading. Collaboration signals — approvals, mentions, assignments — still only came via email or Slack and had no permanent home in the app.
Boost a Post — the first slice of paid amplification ships behind a flag
For Merchants and CSMs who already lean on paid social — especially teams running Meta ads alongside organic campaigns who want to amplify a winning post without context-switching into Ads Manager.
A new Boost a Post side panel lets users pick a published social post and queue a paid boost for it. The first screen lists boost-eligible publications, a prerequisite check verifies the post is actually published and the right ad account is connected before the flow proceeds, and a dynamic config form (driven by feature-flagged field definitions, so we can iterate without re-shipping) collects budget, audience, and duration. When a campaign is set up to be boosted, the Schedule and Publish Now flows now remind the user that a boost is queued so the paid spend doesn't go out by surprise. The whole flow ships behind a feature flag for the first cohort.
Why it matters: When a post performs well organically, the next move is usually to put paid behind it — and that meant leaving Glowtify, opening Meta Ads Manager, finding the post, and rebuilding the audience. Nothing in Glowtify knew the post was being boosted, so scheduling a related campaign happened blind.
Campaign chat got smarter, safer, and less likely to nuke your work
For Everyone using campaign chat as their main editor — operators and merchants iterating on copy + visuals, plus brand managers who care that the guardrails stay in sync with how the AI actually behaves.
Several improvements landed in the campaign chat in the same week. When the AI spots a pattern that should become a reusable guardrail (e.g. "never mention price in captions"), it now suggests adding it as a guardrail directly from chat — with CTAs to review and edit the proposed wording before applying. When the chat generates a new image for a task that already has one, it no longer silently overwrites; you get a confirmation bar with a side-by-side preview to apply or skip — and the bar is smart enough to skip the prompt entirely when the existing image is just a fresh auto-gen with nothing on top. The input stays typable while a response is generating, so you can draft the next prompt without waiting. Image generation no longer inherits the heavy text-generation context blob (SEO keywords, competitor lists, language rules) — the model now gets brand identity, visual guidelines, and only the campaign fields that actually affect a picture. Marketing-event suggestions are anchored to today's date so they stay inside the next 30 days. Bundled fixes: regenerated images in chat no longer fail with "couldn't be saved" on template-based posts.
Why it matters: Earlier, regenerating an image would silently replace the one you'd already styled, a generic ZodError could surface when the chat handled a recovery case, the chat input greyed out for every response (so you sat waiting to type the next thing), and event suggestions sometimes proposed dates that had already passed. Adding a new guardrail meant leaving the chat to find the right settings page.
Multi-Channel Template Sets are on for everyone — and new stores start with one already built
For All workspaces — but especially newly onboarded stores and self-serve trials, where the first campaign experience is the make-or-break moment.
The feature flag gating Multi-Channel Template Sets has been removed, so every workspace can now build and save sets that combine one template per channel (Instagram, LinkedIn, email, etc.) and reuse them across campaigns. New stores no longer start empty: they ship with a default multi-channel template set already populated, so users have something concrete to launch with on day one instead of staring at a blank slate.
Why it matters: Multi-Channel Sets shipped behind a flag last week to a small cohort. Everyone else still had to pick templates one channel at a time, and brand-new stores started with nothing to launch from — which made the first campaign disproportionately painful to set up.
Editor polish across emails, carousels, and templates
For Anyone who spends time in the editor — particularly marketers polishing email copy and operators iterating on multi-page social carousels.
Email tasks now expose the same refine controls (shorter / friendlier / more urgent, etc.) for the subject line and preview text that the body already had. Swapping a single image in a multi-page carousel now replaces just that page in the scene file — the other pages stay exactly as they were instead of getting rebuilt. The Templates page got tighter: an unused section was removed so the remaining options stand out, and replacing a template now shows a confirmation dialog so a misclick can't wipe out a layout you didn't mean to swap. Uploaded post images now have the same edit and delete controls that generated images have, and multi-paragraph AI captions now actually render as multiple paragraphs end-to-end (rather than collapsing into one continuous block).
Why it matters: Refining an email used to mean rewriting the subject and preview yourself even though the body had a one-click refine. Swapping one image in a carousel could subtly re-render the others. Replacing a template was a one-misclick way to lose work, and AI captions designed as multiple paragraphs were arriving as a wall of text.
Cross-app reliability fixes that show up in everyday flows
For Everyone who uses these flows — and especially CSMs / support, who were fielding the resulting tickets.
A handful of small but visible reliability fixes landed: picking a Mailchimp audience segment now returns every segment (not just the first page), so accounts with many segments can actually find the one they want; button links in Beefree-built emails using "send email" now correctly open the mail client instead of corrupting to https://mailto:... and silently navigating to the store's website; revoked Google Ads OAuth grants now return a clean 403 with a reconnect prompt instead of bubbling up as a generic 500; user names now render cleanly when a first or last name is missing (no more "Published by Justin undefined" or empty "Published by"); the post editor's Save and Close CTA reverted back to the familiar X icon close button after it was getting in users' way; the imgly editor now opens single images (blog featured images, Canva cards, social templates) with the image as a selectable, resizable element instead of baked into the page; and the Remix Content power-up's description and layout were tightened so users know it accepts URLs only (no longer hinting at PDFs and images that were never supported).
Why it matters: Each of these was a quiet papercut: a Mailchimp segment you couldn't pick because pagination was hidden; an email button that looked fine but sent customers to the wrong place; a Google Ads error that filed itself as an engineering incident; a published-by-line missing a name; a close button that didn't match the rest of the editor; an image you couldn't move; and a power-up description that didn't match what the system actually did.
