After you add your first video, one results page shows live progress, your full transcript and a free set of show notes, a LinkedIn post, an X thread and a blog post — usually within a couple of minutes — then your clips as soon as they're ready. Take a quick tour or connect your socials while you wait; the tracker in the top bar brings you back when everything's done.
Your first video now gets 30-second clips, which come back faster, and you keep every clip we make from it instead of just the first few. The top three show first; the rest are a click away under "Show more clips".
When Studio finds more clips than your plan includes (3 on Free, 5 on Starter), the extras now appear on your project as locked clips instead of being discarded. Upgrade and they unlock on the same project — no need to process the video again.
The top-bar job tracker now opens by itself when a video finishes (or fails), so you can jump straight to the result. We also fixed a timing bug that could make a stalled video wait much longer than it should before being stopped and refunded.
Clips now default to our more thorough processing mode for everyone — it delivers at least one usable clip far more reliably and with a much lower failure rate than the faster mode, even though total time-to-ready is actually quicker on average. The faster mode is still available for short videos under 2 minutes.
The "Remove Filler Words" option (part of Studio Sound and Podcast Pro) wasn't being applied, so enhanced audio could sound identical to the original. Also smoothed out the playback progress bar, which used to visibly lag and snap during playback.
Fixed a bug where some audiogram videos — most noticeably on Episode Autopilot with multiple clips processing at once — could be paired with the wrong episode's video. Clips also now show cover art on Autopilot.
Describe your show's topic and get an SEO-structured, directory-ready description — hook up front, natural keywords, no stuffing. Free, no signup required. Also available to Enterprise API partners as the show_description tool.
Describe your show's topic and get AI-generated name ideas, each checked in real time against Apple Podcasts or YouTube handles for exact-name collisions. Free, no signup required, no results gated behind a signup wall. Also available to Enterprise API partners as the name_suggestions tool.
Repurpose now generates 10 directory-optimized episode title options (front-loaded keywords, guest+topic formulas, podcast-app length limits) as its own content type, available in the dashboard and via the partner API's repurpose tool.
Personal check-ins from the founder now go only to people who've actually used Atomiser a lot, quote your real usage, and count towards the same limit as all our other emails: at most two a week, and your email preferences always apply.
Two more Atomiser capabilities are now available through the partner API: repurpose (turn transcript text into tweets, LinkedIn posts, show notes, and 12 other formats — billed only for the assets that generate successfully) and publish (end users connect their social accounts via OAuth and schedule posts, metered per destination account). The full integration guide is now public at atomiser.ai/partner-api-guide.md alongside the OpenAPI spec.
On a monthly plan, flip the pricing toggle to Yearly and choose "Switch to yearly" on your current plan. You'll see exactly what you pay today before checkout. Moving from yearly back to monthly waits until your paid year ends, so you never lose time you've paid for. Yearly plans still get their credits every month.
If you've cancelled but your plan hasn't ended yet, you can resume it in one click from Billing or the sidebar. Everything carries on as before and it renews on the usual date.
You can now switch to a lower plan from the pricing page. You keep your current plan and its credits until your renewal date, nothing is charged today, and you can undo the switch any time before then from Billing or Pricing.
If you pay yearly, changing plans now stays on yearly billing by default instead of switching you to monthly. We'll tell you clearly if a change also switches how often you're billed.
On a paid plan, you'll now see an upgrade option under your plan in the sidebar. Choosing a higher plan shows exactly what you'll pay today, what you get and when it renews, then takes you to a secure checkout. You stay on your current plan until the payment goes through, and if you close the checkout you can pick up where you left off.
If you're in India, pricing now defaults to INR and checkout offers UPI (alongside cards). Prices shown include GST, so what you see is what you pay.
The payment confirmation page now checks your actual account status instead of assuming the checkout redirect means the payment went through. If the payment is still processing (common with UPI mandates), you'll see that honestly.
Clicking this suggestion after a clip finishes used to drop you on a blank list of 12 content types with nothing picked. It now opens with Blog Post and Twitter Thread already selected — matching what the suggestion promised — so you can generate right away.
In scenes with two people (or sometimes even one), the camera could drift toward the wrong face for a moment and then jump back. The crop now holds steady on whoever it's tracking and only switches when a different person is clearly the focus for a full second.
AI-generated titles for Studio clips occasionally echoed back the placeholder name instead of a real title. Fixed the underlying transcription step so titles are generated correctly, or fall back to a clean generic title instead of the placeholder.
Pick 30s, 45s, 60s, or 90s with a single tap instead of dragging a slider to an exact second nobody needed. Advanced options are now also available in Studio mode, not just Fast.
When a clip cuts to a text card or B-roll with no one on screen, it now shows the whole frame instead of a tight, hunting crop that clipped the text — the camera returns to tracking the speaker the moment they're back.
Your billing page shows a reset date, but the first free refresh used to arrive weeks later than promised. Credits now refresh exactly on that date each month.
If you cancel a paid plan, you keep full access until the end of the billing period you already paid for, with a confirmation email — instead of being moved to Free immediately.
If a renewal charge fails, you keep your plan for a grace period while we retry the card and email you to update your payment method. Your plan only ends if the payment can't be recovered.
The dismiss control on each Next Steps suggestion only appeared on mouse hover, so it was invisible — and unreachable — on mobile. It's now always tappable.
If a video costs more credits than you have, the upload window now tells you upfront — before any file is uploaded — with a persistent upgrade panel showing the exact math and what each plan gets you. The upgrade state no longer disappears after a few seconds.
As soon as your first clip finishes rendering you can open the project page and start watching — the remaining clips fill in as they complete instead of making you wait for the whole batch.
AI Reframe supports videos up to 30 minutes. For longer recordings like podcasts and webinars, you're pointed to AI Clips, which finds and reframes the best moments automatically.
If a video needs more credits than you have, the upload window now keeps you in place with an amber panel — exact credit math, a View Plans button, and confirmation that nothing was charged — instead of a red error and a dead end.
If a video gets stuck mid-processing and has to be marked as failed, any credits charged for it are now refunded automatically — previously a timeout could fail the project without returning credits.
If our clip-generation provider fails mid-job, your project now automatically retries through our second pipeline instead of showing an error. If nothing can be generated, credits are refunded — you never pay for a failed run.
If a video link can't be fetched (private, region-locked, or platform-blocked), you now get a clear explanation with a direct option to upload the file yourself — instead of a generic failure. Error messages also no longer mention internal service names.
If a video needs more credits than you have, it now shows a clear 'out of credits' state with an upgrade link and the exact credit math — instead of a red 'Processing Failed' error. No credits are charged for blocked videos.
When a tool finishes, suggested follow-ups open with your content already loaded: "Publish to social" opens the composer with your file attached, audio outputs route to the Audiogram maker (which turns them into shareable videos), and Repurpose opens with your transcript or script pre-filled. Suggestions that can't carry your context no longer appear — no more landing on empty screens.
Stale API calls were silently failing, leaving the credit-balance header blank and the plan badge stuck on FREE on three pages. All now point at live endpoints.
The free Meeting Notes tool now produces correctly named files and downloadable results without a gate, and signed-in users can unlock inline instead of hitting a dead end.
The File Converter page now collapses your conversion history by default so the tool itself stays front and center.
Platforms can now embed Atomiser tools (transcription, subtitles, clips, reframe, dubbing, TTS, and more) inside their own products via a REST API at api.atomiser.ai. Jobs are asynchronous with signed webhooks, idempotency keys, cancel/retry, per-end-user usage metering, and test/live API keys. Public docs live at atomiser.ai/developers, with a machine-readable OpenAPI spec for generating clients.
Pasting a direct link to a media file (a URL ending in .mp4, .mp3, .wav, .mov, .ogg, and similar, including presigned cloud-storage links) previously failed during import. Direct files are now downloaded over plain HTTP with redirect support and probed for duration correctly.
Jobs that fail or are cancelled are never billed, and any credits the pipeline had already deducted are now refunded automatically, including dubbing jobs which previously had no refund path.
After finishing clips or transcription, the "Add captions" and "Create clips" suggestions now land you directly on your project with the right tool open — instead of dropping you on a blank dashboard with no context.
Mobile users previously had no persistent way to see pricing or upgrade — the upgrade button was hidden on small screens. It now appears in the top bar on every page, alongside your trial countdown if you're on a trial.
Fixed a bug where dubbing jobs longer than a minute were being charged as if they were exactly 60 seconds. Credits now reflect the real length of your source file, and Start Dubbing is disabled with a clear message if your balance can't cover the job.
The Audiogram tool now correctly rejects video files (it's audio-only) instead of silently accepting them, and no longer gets stuck showing "Creating..." after your first attempt.
Converting a file no longer redirects you away immediately — it now shows up in the same background job tracker used by AI Clips, Subtitles, and Repurpose, so you can keep working while it finishes.
Importing from a URL in Visual Assets is now billed at the standard transcription rate (0.25 credit/min) instead of the higher Clips rate, and its transcript auto-populates from any linked project. Removed Upload Image and Stock Photos as background sources since they couldn't produce a transcript for text-based asset types.
The Cancel Subscription action now reliably reaches your billing gateway, and we removed a "Manage Billing" button that wasn't wired up to anything.
AI Audio Enhance now opens as a modal (matching AI Clips and AI Subtitles) and tracks progress in the background job tracker, so you can keep working elsewhere while it processes.
The Transcription modal now supports importing directly from a URL, and a new /transcribe history page lets you revisit past transcripts with a switcher to jump between them.
Fixed the clips project page briefly showing "no exports yet" and phantom duplicate clip cards while loading. Also added a clear fallback if auto-render ever stalls instead of an endless spinner.
Corrected AI Reframe's description to reflect real AI face tracking, rather than implying a static center crop.
Smart Crop now zooms in based on how close or far a speaker is from the camera, so distant subjects are framed as a proper close-up instead of a tiny figure lost in a wide shot. Vertical clips also anchor faces higher in frame for natural headroom, and two-person shots keep both speakers visible instead of pushing one out.
Added a tiny audio fade at every clip cut point to eliminate the audible click or pop that could happen when a cut landed off a clean point in the waveform.
Studio mode (higher-production AI Clips) now offers a style picker with 8 looks — the default is free, the other 7 (Apple, Black Box, Pop of Yellow, Pink Pony, Conspiracy, Basic, Blue n Bold) are available on PRO+.
If a post you scheduled to social media fails to go out, you now get an email explaining what happened, which platform(s) failed, and confirming your credits were refunded — with a direct link to check your connected accounts.
When your AI-generated clips finish rendering, you now see personalized suggestions for what to do next — add captions, create thumbnails, repurpose into posts, or dub into another language. This helps you discover the full creator pipeline instead of leaving after clips.
When you run out of credits and try to export or enhance, you're now taken directly to pricing instead of seeing a dead-end error. The upgrade modal also shows more frequently (every 8h instead of 24h) and uses clearer copy that prioritizes action over waiting.
AI Repurpose used to require a transcript from another tool (AI Clips, Transcription) before you could use it. Now you can upload a video/audio file, paste a URL (YouTube, Vimeo, TikTok, X, Rumble), or pick from your Asset Library directly from the AI Repurpose card on your dashboard — we transcribe it in the background (same rate as standalone Transcription, 0.25 credit/min) and bring you back once it's ready to turn into posts, blogs, and more.
A bell icon in the top bar now shows new changelog entries and admin announcements. Changelog items auto-populate from the same JSON the CI check enforces, so the bell is always current without manual upkeep. Admins can also create one-off notifications from the admin panel for time-sensitive announcements.
The changelog now has an RSS feed at /changelog/rss.xml for users who want to subscribe via feed readers. Reads from the same auto-updating JSON as the page.
Users with product update emails enabled now receive a monthly summary of what shipped. Reaches you even if you don't log in — part of our effort to make sure no one misses improvements.
Admin feature-announcement emails previously bypassed opt-outs, bounce suppression, and internal-account filters. They now go through the same safety checks as every other email — if you turned off product-update emails, you will not receive them, and bounced addresses are no longer re-emailed.
A permanent, public record of what changed on Atomiser. Linked from the in-app help menu so you can check what shipped without waiting for an email.
The 24-hour project-expiry warning was sending but not appearing in our email dashboard, and a user with multiple expiring projects could receive duplicate emails. Both are fixed — warnings are now deduped per user and visible in our logs.
When you import a video, Atomiser now checks for an audio track before processing starts and warns you if none is detected — so you do not spend credits on a file that cannot be transcribed or clipped.
After finishing a tool, Atomiser now suggests a logical next step (e.g. after transcription → generate captions). The surface respects a per-suggestion dismissal, a 7-day per-source cooldown, and a 4-hour global cooldown so it never repeats the same suggestion twice.
The upgrade prompt that appears when credits run low was displaying the wrong price in some cases. It now shows the correct Starter and Pro pricing.
Email suppression was not checking the bounce/complaint flag, so a bounced address could keep receiving emails. Suppression now blocks any address flagged by Resend webhooks.
A repeated audiogram completion callback could trigger multiple emails to the same user. Processing now skips if the audiogram is already completed.
When subtitle export failed, credits were sometimes not returned. Failed exports now refund the credits used.
The win-back email previously used a field that changed on unrelated database updates, causing it to target the wrong users. It now targets users who were genuinely active 8–14 days ago based on actual credit usage.
Lifecycle emails suggesting a next tool used to link to the generic dashboard. They now link to the specific tool page with UTM parameters so we can measure which suggestions work.
The free tools (transcript generator, audio-to-text, MP4 to MP3, SRT subtitles, show notes) were previously unlinked from the rest of the site. They are now linked from the homepage, blog, and persona-page footers so visitors and search engines can find them. Also corrected the tools pages to say 30 free credits/month and 5 free runs per day, matching current limits.
The free-tools hub and tool pages were polished to match our design language: pastel feature cards on the hub (same style as the homepage), the brand teal accent instead of off-palette green, and a proper footer with site links on every tool page. All nine free tools are now linked from every footer.
We've kept a private build journal since day one — ~200 entries across ten months. Now we're publishing it on the blog, one journal day at a time, lightly edited for privacy. Entry 1 covers launch day: shipping the MVP and immediately rebuilding its own UI.
The footer is now organized into clear sections — Product, Free Tools, Resources, and Legal — with our logo and social links up top, replacing the single flat row of links. It now appears consistently across the homepage, blog, persona pages, and every free tool.
Every project card on your dashboard now uses the same hover menu (the ⋮ button with Rename/Delete) instead of two different icon styles. Dubbing, Transcription, Text to Speech, Studio, and Workflows now have the same back button as the other tools. Audio Enhance also switched to its proper indigo accent color.
Have feedback or a bug to report? Email contact@atomiser.ai.