Every release from the last 90 days, newest first. RSS
You can now describe your video right on the FinalTake home page. After you sign up, what you typed is already waiting in the studio.
Buttons, loading spinners and confirmation windows now use FinalTake blue everywhere, not only inside the studio. Some pages had kept an old gold colour.
The projects list view now has a checkbox on every row, so you can select several projects and delete them together, just like in the card view.
Your projects open about twice as fast and more of them show a thumbnail. Tick several projects to delete them together. Every “are you sure?” now appears inside FinalTake instead of a browser pop-up, and the brand switcher shows your logo.
A storyboard warning now disappears as soon as it no longer applies, instead of waiting for your next edit. And you can close any warning with the ×.
Premium lip sync and the expressive presenter are now one choice on the storyboard: Basic, Premium, or Premium + gestures. And on a talking shot that is already made, one button re-shoots it walking.
Walk & talk is now open to every account with Premium lip sync: your presenter can walk toward the camera while saying their line. And every talking shot is now checked for lip sync; if the picture runs ahead of or behind the words, it is moved back into time automatically.
With Premium lip sync on, you can make your presenter expressive: they are framed from the waist up with their hands in the shot and gesture naturally as they talk, at the same price. Some accounts can also try walk & talk shots, where the presenter walks toward the camera while speaking.
Pressing Render on a video you have already rendered, with nothing changed, now brings back the finished video straight away instead of rendering it a second time.
On the Final page, pick up to five languages and get the same video in each: a new voice, captions, name cards and on-screen text in that language. Spanish, Portuguese (Brazil), French, German, Italian and Arabic.
On the Final page, pick another presenter from your library and get the same video with them on camera. The script, b-roll and music stay the same. By default the new presenter speaks in their own voice, or you can keep the current voice.
Hover over a stock presenter, or long-press on a phone, to see a bigger photo and hear them introduce themselves. Several presenters now speak with an accent that fits them, and a voice you choose yourself stays chosen.
After we break down a video you want to clone, you can save it as a template and start from it again any time from the template library.
The template library has eight new starting points, from Problem → Solution to a faceless listicle. A template now also sets the color look and caption style, and the plan tells you whether it kept the template’s shape.
The automatic review of a captioned video no longer reports a word as cut off just because it is halfway through lighting up. Captions are now checked against where they are actually placed, and square videos keep their captions clear of the app bars.
A name card now sizes its text to fit the frame. On vertical videos, even a short name like “Anna Reyes, founder” used to lose its first and last letters off the edges.
A name card added to a scene now sits above your captions, so neither covers the other.
Videos made on the free plan now carry a “Made with FinalTake” mark. Paid plans still carry no FinalTake mark at all.
Turn on comments for your share page and anyone with the link can pause your video and leave a note pinned to that second, with just their name and email. You get an email for each one and can resolve them from the Review card.
The plan now appears as soon as the Director has written it, while the crew does a last read-through of the finished board.
When the plan looks right, one button makes it, and the button shows what it costs.
Presenters in 1080p videos now render at full 1080p instead of a smaller size scaled up. It costs more credits, and the quote shows the higher price before you approve.
A caption word that sits behind the person on screen now includes the cost of cutting them out in the price you see before you generate.
A new Developers page lists everything the FinalTake connector can do and gives copy-paste steps for Claude, Cursor and ChatGPT.
A new Account tab in Settings lets you set or change your password, email yourself a reset link, and delete your account and personal data.
The changelog now reads like an edit: every release is a clip on a timeline you can scrub, here and in Settings.
After each turn the transcript lists what the Director actually changed, in plain words.
Every release that changes what you see is now listed publicly at finaltake.ai/changelog, with an RSS feed.
Connecting Claude, Cursor, Codex or your own code to FinalTake now needs the Creator plan or higher. Free accounts see why on the connect screen and can upgrade from there.
The finishing tools now show which plan they come with, and the free plan can see them but not spend on them. Nothing you already had access to has moved down.
The presenter picker had two stock faces. It now has forty-two — a range of ages, looks and settings, each paired with a voice — so a video with nobody to film still gets a presenter that fits the product.
On a finished video, ask for five alternative opening lines, pick the ones you like, and get a cut for each. Only the opening shot is re-made — every other shot is reused — so a variant costs one scene plus a render, not a whole video.
Every finished video now has a page you can send to anyone — no login, just the video, its title and a way for them to make their own.
Closing the tab while your shots were being made used to leave the project stuck at "every clip done" with no video — the final cut only ever started from an open page. Now the studio cuts it for you and tells you when it is ready.
TikTok has approved FinalTake for public posting. The TikTok channel is now open to every account, and posts can go out as public.
The starter cards on the home screen now show the kind of video each one makes — a vertical reel with the captions, split screens and beat bars you would actually get — and templates come with moving previews.
Real motion on any shot, a new video type for property photos, one-tap templates, 4K export and music made for your video — all inside the same plan-then-approve flow.
A new "Premium lip sync" switch on the storyboard gives talking and singing shots sharper, more accurate mouth movement, for more credits.
In real-estate videos, your presenter now speaks with clearer mouth movement, in a sharp close portrait placed in your real location. Those talking shots cost a little more, and the price you see before generating includes it. Other video types are unchanged for now.
Property launches get their own product type. A brochure or photo of a building is recognised as real estate, the script is written as a property launch, and no shot tries to put the building in your hands.
A permit number and broker number together were running to the very edges of the picture. The line is now sized to sit comfortably inside the frame.
You can now post straight to Instagram. It goes up as a Reel, so FinalTake checks the video is vertical first — a wide one would post and then be hidden from the Reels feed.
Change a finished project to Arabic and the captions change lettering to match — and change back again if you switch to English.
Two things in the Publish panel were telling you the wrong story — a privacy setting that did not apply to the channel you picked, and a TikTok refusal that handed you a link instead of a reason.
Save your advertising permit number once and it appears on every shot, with your permit QR code on the end card where someone has time to scan it.
Drop in a PDF brochure and FinalTake now keeps the pictures inside it, not just the words. The renderings, floorplans and room shots become the project’s photos, and each shot in the video is built from a different one.
TikTok joins YouTube as a channel you can post to from the Publish panel. Because TikTok asks that you make every choice yourself, the panel now reads your account live and shows you exactly what it will do before anything leaves FinalTake.
Paste a TikTok link on the start screen and FinalTake watches it for you — what its opener does, how fast it cuts, what it sounds like — then fills in a brief you can change before anything is made or charged.
When you upload an episode, the choice between using part of it and using all of it is now a single pick, and the button tells you how big the video you are about to make really is.
An uploaded episode longer than one video now becomes several, made one after another and joined at the end — so a twenty-minute show is no longer something you can only take the first few minutes of. Being rolled out gradually.
Turning an uploaded episode into a plan used to happen first and mention the price afterwards, so you could sit through the whole thing only to be told at the last step that it was never affordable. Now you hear it up front, while changing the range is still easy.
You can still upload a long episode — that has always been the cheap way to work, because you pull a short clip out of it. What changed is that "Full episode" no longer means "all nineteen minutes of it", which used to build a video far larger than any plan could pay for.
Some moments are better performed than narrated — a dance, a turn, a reveal, walking in and landing a pose. Your presenter can now do those as a proper sequence, planned out shot by shot so each one picks up exactly where the last ended and the whole move reads as one continuous take.
A scene shot from an angle was coming out almost in profile — your presenter turned so far you were looking at the side of their face. The turn is much gentler now, so an angled scene reads as someone turning toward what they are talking about rather than turning away from you.
Choreographed shots were drawn from a character sheet, and a sheet fits three whole figures into one picture — so the face on it was tiny, and every shot drawn from it inherited a fainter version of you. They now start from your photograph instead, and each drawing is checked before anything is filmed.
Your presenter is photographed once from each angle and those photos are reused in every video. Some of the angled ones were not the angle they were meant to be — one came back as a full side view — so cutting between two shots of the same angle made the head jump. They are now checked and re-shot when they are wrong.
The Director can now write shots where your presenter MOVES instead of talks — a dance step, a turn, a product held up on the beat — as a choreography board that is drawn before anything is filmed. On for test accounts only while we look at the results.
One house caption style — Pop, size M, lower third — instead of a look chosen for you that often landed across your face.
It now knows which shots are filmed, whose mouth has stopped matching its line, and what the quality check found — and it tells you before an edit costs you a re-shoot.
Right answers, delivered like a colleague rather than a build log — and an end card no longer inherits a line the video already said.
Telling the Director to change something is a conversation now — it answers, you can reply, and it says which shots need re-filming and what that costs before you spend.
Changing a line re-recorded the voice but kept the old footage, so the mouth no longer matched. Rendering now stops and tells you which shots to re-make.
The first split frame rendered with two thirds of its half left black, because the card was drawn for a whole frame and then shrunk into it.
Changing the look after the crew had planned your video silently kept the old look. Clarify now tells you, offers to re-do the crew’s notes, and names the button you need.
Opening the storyboard used to start making the video on its own, credits and all. Nothing is made until you press the button. Split frames also work on vertical videos now.
Pulling pictures from a web page returned broken images on most modern sites, and offered you logos instead of the product.
Opening Storyboard on a finished video sent you straight to Final Cut instead. Also: graphics now really do sit beside the presenter when you ask them to, name cards name the person on screen, and the chat can take a page address or a logo.
Unused monthly credits no longer disappear at the end of the cycle, the free plan now starts with 100 credits, and Studio gives 4,000 for the same $249.
Two paid plans advertised things the product does not do. The pricing page now lists only what your plan really gives you.
A before/after graphic in a vertical video used to run off both sides of the frame, slicing the first and last letters off each side. It now stacks.
The marketing site now says what we make and how we differ from Captions: directed ads you approve before spend, not an editor for footage you already shot.
Videos with background music or room tone could end up to a second short, cutting off the end of the final line. They now always run the full length of what was said.
Captions can now sit behind the person on screen instead of on top of them — the whole track, not just one word — so they read as part of the shot rather than something stuck over it.
Captions were one style with the paint changed. Now there are fifteen real looks — solid bars, printed cards, spaced caps, two words at a time — and every one is measured against the frame before it is burned in.
Everything the video is made of — footage, music, sound effects, your own files — now sits in one list on the Final page, with where it came from and what it obliges you to credit.
Text cards, stat callouts and comparisons now come in ten looks across four layouts, and the Director can reach for them by name instead of falling back to a bullet list.
Cut, crossfade and dip to black now have company: whip pan, flash, clock wipe, pixel dissolve, blinds and push in.
Blockbuster, Bleach and Faded join the grade picker, and every look now renders through a proper film lookup rather than a stack of sliders.
Every on-screen graphic is now checked for text running off the edge, elements landing on top of each other, and colour too faint to read — before the render starts, not after you have paid for it.
A backing track now makes room for the narrator instead of competing with them. The bass and the air stay where they were, so the track still sounds like music rather than going flat under the whole video.
Planning a video now asks one more thing: when a number or a chart matters, should it sit beside your presenter or get a shot of its own? You pick from two pictures of the frame, and nothing composes a shot unless you said it could.
A shot can now seat your presenter in half the frame with a graphic building beside them, instead of a card sitting on top of the footage. Good for the one moment in a video where a number deserves its own half of the screen.
When a script came out meaningfully longer than the runtime you picked, the video simply ran over. It now reads a little faster to pull back toward the length you wanted — the same words, slightly brisker, rather than a video that overruns or a script you have to cut.
Explainers can now draw a timeline and a breakdown. Both were shapes the graphics could not make before, so a history or a "where it all goes" slide had to fall back to stock-looking footage that cannot spell a year or hold a number.
A new look for the graphics in your videos — frosted, translucent cards floating over drifting colour, the way a phone or a smart watch draws them. Pick it once and every graphic in the video uses it.
If a cutaway is not what you pictured — or you already have the real thing on your phone — you can drop your own video into that shot. Your voiceover keeps playing over it, and the shot stops costing credits.
A lower third, stat callout or any other motion graphic laid over a shot used to appear in the editor as a dashed box reading “in the final render”. It now plays where it will actually land, so you can see it before paying for a render.
A video with a presenter can now cut away to somewhere genuinely else — a desert, a skyline, a different city — without the shot being told it is still standing where the presenter is. Shots that share the presenter’s room still hold it.
Until now every automatic check looked at one shot at a time, so a video where the presenter changed outfit or an object changed shape across a cut passed them all. The review now compares shots against each other, and against what the plan said should stay the same.
The backdrop question now offers eight rooms to choose from — including two outdoors — so a good-looking set is one tap. Typing your own is still there if none of them fit.
A conversation can now play as a grid — every person visible for the whole video, the speaker lit, and the others listening rather than frozen — instead of cutting to one face at a time. Available on every account.
On angled shots the presenter faced the camera but their eyes slid off to one side, which reads as shifty. The head now actually turns, and the eyes go with it.
The warnings on the storyboard now use the words of whatever you are selling. A phone video is told about the device, its case and its screen — not about a bottle, a dropper and a droplet on the cheek.
A phone explainer was warned that a stranger was in the shot and that product might appear on skin before it was applied. Neither was true. The checks were written against beauty videos and read every other kind of video in that vocabulary.
When a video starts from a product page, the film that page publishes about the product is now offered shot by shot — the maker’s own footage of the actual thing, instead of a generated approximation of it.
If we have a photograph of what you are selling, every shot is built from that photograph instead of from a description someone wrote of it.
Ask the Director for a picture and it pulls candidates from your product’s own page, or from free stock, and shows them. You pick one; nothing changes until you do.
Layers used to be something the crew avoided unless you asked. Most of them are free, so avoiding them bought nothing and every video came out as a run of full-frame shots. Now a beat that is richer with a name, a mark or your product photo in the corner gets one.
Layers had four number boxes — X, Y, width, height. Now there is a picture of your shot: drag a layer where you want it, grab a corner to resize. And a layer no longer moves between the editor and the finished video.
The Director kept asking for a product photo in a box you could only type into. There is a paperclip now, and what you attach is filed where the render actually looks for it.
"Change something" now opens a box you type into rather than a form of presets, and it keeps going until the plan is right. The brand question finally lets you say the video is not for one.
An explainer is carried by what you can read, not by stock footage. The crew now plans one graphic every 8–12 seconds on explainers and faceless videos, and a graphic scene costs a flat 5 credits instead of being priced like AI footage.
Steps and diagrams. A sequence and a relationship are what "how it works" is made of, and until now both had to fall back to a bullet list or to AI footage, which cannot render readable text.
Paste a link and we now read the page itself, not just its title. Attach a PDF, a text file or notes and the crew writes from them. And a link no longer decides that your video is a product ad.
A podcast used to be exactly two people. Now up to four can be at the table, and everyone at it is put in the same room — so three people who recorded in three different countries cut together as one conversation.
The plans no longer guess how many videos you will make. Instead the pricing page prices real deliverables — a 15-second reel, a 30-second ad — so you can work out your own month.
A portrait photo used in a widescreen video no longer loses the presenter’s chin, and the planning chat now asks where to set the scene before the crew starts.
A photo with someone else in the background used to make them part of your presenter for good. Now that background gets replaced by default, the result is checked, and your script’s own setting is one tap away on the plan.
The plan you approve now shows a picture for every beat instead of a list of sentences, so you can tell at a glance what the video will actually look like.
The biggest change to graphics so far — eleven new types, a look that follows your brand instead of always being dark blue, and an end to the still frame every graphic used to hold.
We advertised 150 free credits and handed out 60. The number on the site is now the number in your account.
Shots were changing between beats instead of on them — the one thing that makes an edit feel like a music video.
Videos built by an assistant connected to FinalTake came out with no music bed and no subtitles, because there was no way to ask for either. Now there is.
Videos can now finish on an end card — your logo animated on a brand-coloured card, or a closing video of your own. And picking a logo for a shot no longer means pasting a link.
Five things that were in testing are now how the studio works — wider and tighter shots, a crew that argues about framing, and more.
The director now knows which part of your track is the biggest moment, and the screen says so when something it tried to do did not work.
The prompt-issue warnings now come with a button that fixes them, and one warning that could never be cleared is cleared.
Performance shots in a music video can now render on a model that actually performs, instead of one built to hold a presenter still.
The new video model had started adding its own soundtrack to b-roll shots. It is now asked for silence, and told what not to draw.
Uploading a photo now tells you which kinds of shot it can give you, at the one moment that is easy to change.
Two new caption looks, and a starting choice that follows what you are making rather than one house style.
You can now say how your video was shot — polished, plain, or like something filmed on a phone.
Your presenter is no longer framed the same way in every single shot. A beat that wants to land closer now gets a closer shot.
The crew reads what your presenter is doing with their hands, and says so when a beat needs a wider shot than the one it is getting.
B-roll shots are now described to the video model in its own language — shorter, with the subject first and the camera stated. Several things that could only confuse it are gone.
Every b-roll shot now renders on a newer video model. Sharper output on 1080p projects, and the quality presets finally mean what they say.
The free plan has always said “up to 720p”. It now renders at 720p — and your credits go further because of it.
The project tools no longer sit greyed out in the sidebar when you have no project open, and Settings is finally usable on a phone.
Briefing the crew from a connected assistant used to go quiet for minutes and then tell you to check back. Now you see each part of the crew as it starts, and the finished storyboard arrives in the same breath.
A video started from a connected AI assistant was quietly built by an older, simpler crew than the one the website uses. It now runs the same crew, and it can show you what it made.
When you connect FinalTake to Claude or another AI assistant, it can now see your brands and products and take photos from you — things it simply could not do before.
Picking your voice could answer with the word “Unauthorized”. Your sign-in had quietly run out, and nothing renewed it.
Presenter clips are now generated at a higher resolution, and they start rendering sooner when the studio is busy.
The voice had started adding words of its own — an “uh” here, a half-repeated word there, none of them in your script. It no longer does.
Your presenter now stands somewhere you picked, and stays there for the whole video.
Each angle on the plan had a redo. The master shot they are all built from did not.
On the plan, an angle whose first attempt was rejected could sit on “Not yet” for good, and one that took a while could spin after it had finished. And a presenter could arrive with a head too big for the body.
The plan used to show your presenter on set from one angle. The video uses two or three.
A real shoot blocks the presenter once and moves the camera around them. The studio now does the same: the shot you approve on the plan is the setup, and every other angle is that shot with a different take.
Before anything renders, your storyboard is checked for defects the video model would take literally. Two of them have the same answer every single time, and the app was paying the AI to write that answer out again on every board.
The plan shows your presenter on the crew’s set before anything is spent, with a button to change it. Typing a new set used to change the note on the plan but not the room that got built — the crew’s original description still won. It no longer does, and you get one-tap alternatives instead of a blank box.
The crew locks a production bible before it writes a single shot — one world, one cast, one set of rules the whole film obeys. It had grown long enough to run out of room mid-sentence, and a bible cut off mid-sentence is thrown away entirely, so some films were built with no bible at all.
While your crew builds the plan there are long stretches where nothing streams — the handoffs between the Director and the specialists. The screen used to show three dots, or nothing at all, on a wait you are paying for. It now says what is actually being worked on.
Your narration was recorded one shot at a time and glued together, so every pause between lines was identical and the pace never moved. It can now be read as one continuous take and cut back apart, which is what makes it sound like a person.
The likeness check was rejecting shots that were plainly you, because it had been told that a slimmer face means a different person — and every image model slims a face. It now looks at whether the face underneath has actually changed.
Scripts were coming back flat — a line that continued into the next shot closed itself off, a list of cities read as three separate announcements, and a closing call to action landed at the same pitch as everything else. The crew now writes for a mouth rather than a page.
Cutaways were being written around “the presenter’s hand” and coming back with a whole face — drawn from a description, not your photograph, so it was never quite you. Cutaways now stay on hands, objects and rooms, and other people are welcome in them.
Fixing a warning on your storyboard could quietly leave a different one in its place, and all you saw was a spinner. It now keeps going until the board is clean, and tells you what it changed on the way.
A shot could put a second person next to you — face and all — and pass every check, because the check stopped as soon as it found ONE person it recognised.
The set your presenter is placed into is now checked before anyone is composited into it, an empty room is no longer shown as if it were a preview of you, and generating asks once if you have never actually seen the shot.
A shot could place your presenter mid-air — torso above a desk, legs simply missing, no shadow — and pass the check that is meant to catch exactly that. The check only ever looked at the head.
Cutaways were arriving with no voice-over and the board offered to write one afterwards. Whether the voice runs on over a cutaway is now part of the plan, and your edit style decides it.
The on-set preview was often blank because nobody had ordered the presenter’s angle shots yet. They now start while you are still reading the plan — and every angle magnifies under your cursor, so you can actually tell it is you.
A new Edit style choice on the plan — Fast social, Considered, Brand film or Documentary. It sets shot length, line length, how fast the hook lands and how hard the ending is, so a TikTok comes back cut like a TikTok.
Three fixes to how your presenter is shot: a scene never ships a face the identity check rejected, no scene is filmed from a pure side-on profile, and every video starts on the Clean grade instead of a look the Director picked for you.
Setup no longer stalls after you hand the plan to the Director. One tap starts the crew round, and when the button genuinely is not ready yet, it looks and says so.
Presenters are now framed at a natural conversational distance instead of a tight close-up, so their delivery reads calm and composed rather than over-animated.
Starting from a template used to jump straight to length, shape and voice — and then build a video without ever asking what it was for. Now the subject is the first question, with AI-suggested ideas one tap away if you are still deciding.
The soundtrack picker now searches a library of thousands of Creative Commons tracks — with a weekly Trending chart and TikTok-style genres like Phonk and Lo-fi — instead of offering just three built-in tracks. Every track is cleared for commercial use; preview it under your voiceover before you commit.
The start screen makes room for real scripts, leads with four ideas instead of eight, and can now make a whole video without a presenter — narrated over b-roll and motion graphics. And the Director asks you the few things that matter before the crew starts work.
The plan now shows a photo of your presenter standing in the set the crew chose — the actual first shot of your video, taken while you read the plan. If the set is not right, change it in a sentence, or keep your own photo exactly as you uploaded it. Until now the first look at any of this came after the video was generated.
A recent change made presenter scenes build from the raw photo crop, and it showed: heads cut at the chin floating over bare empty rooms, the odd scene with two of you in it. Scenes now build from your best, checked angle shot, the set behind you is a rich, lived-in space instead of a blank room, and every finished frame is inspected — one person, whole head in frame, right size — before it ships.
The last step on a presenter shot is a quick trim, done by a machine that has to wake up first. If it was still waking, we treated the shot as failed and binned it — even though the shot itself had finished rendering. Now we wait for the trim to land and pick it up where it left off.
When your scene had its own setting, every camera angle was building that room again from scratch — so the furniture, the walls and the light quietly changed each time the angle cut. The room is now built once, like a real set, and every angle is filmed in it.
Cutaway shots often have no line, and staring at an empty box is no help when you do not know what belongs there. Ask and we will write one for each silent shot, reading the lines either side of it so it joins them up. You see every line before it is used, and can change any of it.
If you uploaded a photo taken from a distance, the presenter in your video could come out looking like a different person entirely — the right hair and the right outfit, but not your face. We now zoom in on your photo before anything else happens, and we check every angle we make against the original.
When a scene had its own setting, your presenter was re-photographed into that room once for every single scene — so their face and their clothes shifted a little from shot to shot, and now and then one shot came out plainly wrong. The setting is now built once and shared by every scene that uses it.
The longest video you could make was 90 seconds, and asking for anything longer quietly came back as 90 seconds or less with nothing saying so. You can now make videos of 3, 5 or 10 minutes, and a long film is written a section at a time so the whole arc gets the attention a single pass could not give it.
Choosing a presenter and a brand on the start screen only stuck if you started from a typed idea. Paste a link, paste a script, or tap one of the template cards and your choices were quietly dropped — so the crew cast one of our stock presenters, in their voice, and wrote the script as if your brand did not exist. Every way of starting a project now keeps what you picked.
Most ElevenLabs keys are created with limited permissions, and we were checking for one the app never actually uses — so perfectly good keys were turned away as invalid. We now check the right permission, and if a key is refused we show you ElevenLabs’ own reason instead of a dead end.
Clearing a shot’s script used to quietly throw away the voice-over already recorded for it, and a video with no voice left on it could get stuck rebuilding its audio forever. Both are fixed, and the automatic shot review now tells you when it thinks a clip came out wrong.
If your presenter already has a voice attached, that is the voice you get. The director can no longer quietly cast a different one while planning the video.
The longest wait on the planning screen now narrates itself: each scene appears in the transcript the moment your director writes it, and the production bible can no longer wander off into prose instead of locking.
Approving the plan now opens the storyboard instead of charging you — you see every shot first, and nothing is spent until you press Generate. And shots that are pure scene-setting can use free stock footage instead of being generated, so they cost nothing at all.
Briefing the crew from a connected AI tool now answers straight away and reports back as it goes, instead of going quiet for minutes while the storyboard is written. And when a tool is briefing your crew, the app tells you.
FinalTake now plugs straight into the AI tools you already work in. Connect one from Settings and it can start, brief and finish a video for you without you opening the app.
A cutaway of hands on a product is no longer treated as you, so it does not start as your presenter and turn into someone else halfway through.
The round where your crew argues out the plan used to disappear the moment you left the screen. It is now saved with the project and can be re-read any time.
Until now, a note you gave on one video was a note you gave again on the next. You can now mark any note as something to apply to every video for that brand.
Until now we checked each clip on its own, and never the finished video. Every render is now looked at as a whole, and anything worth a second look is listed under the player.
The review that reads your storyboard for problems only ever looked at what each shot says. It now also checks the timing of anything layered on top.
On vertical videos, TikTok, Reels and Shorts draw their own buttons and captions over the top and bottom of your frame. Anything we placed there was invisible. Now it moves.
B-roll shots that show the presenter — a hand on the keyboard, a shoulder in an over-the-shoulder — now use the same person as the talking-head scenes instead of a lookalike.
Rendering a finished board could fail with nothing to explain it. The repair work it was doing first now happens earlier, out of the way.
A question from the director now comes with a box to answer it in. And when a build simply falls over, it says so and gives your credits back instead of asking for detail you never owed.
Put your brand logo in the corner of every shot from Look & feel on the Editor. It costs nothing and needs no shot regenerated — pick a corner and render.
Regenerating one shot now costs you one shot. When something does go wrong, you get a sentence you can act on rather than the raw text underneath.
Creating a video now starts with a single description instead of a form. You get a plan you can read and argue with before anything is generated — and before you spend anything.
