Change the outfit, place, product or person in your video.

No, and yes. Riffkit never re-uploads the source clip: every frame of your video is newly generated. Adapt keeps only the structure (the hook, the pacing, the beats) and builds new footage around your story, product and character, so nothing of the original's visuals or sound ends up in your video. Swap keeps the original's shots and timing and changes what you name, so use it on your own videos or a template. Don't copy. Riff.

Adapt when the winner belongs to someone else, or when you want a new story: it keeps the formula and generates new footage around your product. Swap when the video is yours or comes from a template and you like its shots as they are: the camera, cuts and timing stay, and you change the outfit, the person, the setting, the product, or the whole look by describing it. Both bill by the second at their own rates, shown before you render.

Yes, that is what Swap does. Upload your video or pick a template, then say what changes: write the new outfit or the new setting in a sentence, or pick a character to take the place of the person on screen. The shots, cuts, timing and music stay where the original has them, and you see the price before you render. Small details can drift, so watch the result before you post it.

Yes, AI-made video runs as organic posts and paid ads on both platforms. The one requirement: if a video features realistic AI-generated people, tag it with the platform’s built-in AI-content label when you upload. Every riff is original footage you own, so it goes through the same ad review as any other creative. What platforms actually act on is undisclosed synthetic content and low-effort bulk posting: label your riffs, vary them, and do not post twenty near-identical cuts. TikTok sells its own AI video generator to advertisers, so AI creative itself is not what gets flagged.

Yes. Riffs render 9:16 by default and post natively to TikTok, Instagram Reels, and YouTube Shorts; you can also pick 1:1, 4:5, 3:4 or landscape ratios at generation. To start from a Reel or a Short, upload the video file; paste-a-link input supports TikTok URLs.

Yes. Reframe any finished video into 3:4, 1:1 or 4:5 for Feed and other placements: near frame-identical, only the framing shifts. Each ratio is its own render, billed per second at the reframe rate shown on the Pricing page.

Yes, that's the point. Add your product image, whether it's a physical item or a screen from your app, and Riffkit weaves it into the story the formula calls for, instead of slapping it onto a generic stock avatar. The winning emotional journey gets rebuilt around what you're actually selling.

A consistent character you define, and it can be you. Upload a reference photo, your own face or any avatar you like, write a persona, and that same character stars across every riff, so the channel keeps one recognizable face and voice. Prefer not to manage a character at all? Auto mode casts one for you. In Swap, the character you pick takes the place of the person on screen; pick none and the person in your source stays.

Build a character in Riffkit: upload a photo (one image with a close-up of the face next to a full-body shot works best, and one clear front-facing photo is enough), write a few lines of persona, and add a 4 to 15 second voice sample if you want a voice of its own. Then pick it for every video. Swap puts it into a video you already have, Adapt tells its own story on a winning video's beats, and Create makes a new video from an idea, so the same face carries every post. Our own digital-human TikTok accounts run this way: of the 1,504 AI videos we tracked, the best passed 1.5 million views.

Yes. Add a short voice sample (4 to 15 seconds of clean speech) to a character, and when you Adapt or create a video, that character speaks in your voice, which is how a personal-brand account keeps one recognizable voice across every video and across every language you post in.

Lock the same character and product across riffs and the series keeps one recognizable face, voice, and cast, while each video’s footage is generated fresh. Consistent enough to read as one channel, varied enough not to repeat itself.

Yes, and it's the fastest way to find out who actually buys. Create a character for each audience you want to test (a runner, a new mom, someone managing fatigue) and pair each with a creative direction that speaks to that buyer's reason to purchase. Riff the same winning formula across all of them and you get a batch of audience-angle tests in an afternoon, where the traditional route means recruiting a new creator for every audience, a few weeks each. One persona for a consistent channel, many personas for testing: same character system, two jobs.

Yes. Riffkit generates natively in 9 languages: English, Spanish, Portuguese, Indonesian, German, French, Italian, Japanese, and Mandarin. Each output is generated in-language, with native phrasing, the right fonts, and captions aligned to the spoken audio. Real localization, not a translated caption layered on top.

Yes, captions are burned into the video and track the voiceover, so the words land with the audio instead of drifting the way auto-captions usually do.

Yes. Riffkit installs as a skill your AI assistant can use: connect it once, then tell your agent to riff a video and it drives the whole pipeline for you. Same engine and the same post-ready output as the app.

Yes. Steer each riff with a creative direction, swap the product or character, or re-run with a new angle. Because a riff is billed by the second, not at the price of a paid re-shoot, dialing in the version you want is fast and cheap, so you re-roll until it's right rather than waiting days on a redo.

As many as you want: the formula is reusable. Re-riff the same source with different creative directions, products, or characters, and each run generates fresh footage, so one proven structure becomes a batch of distinct videos instead of the same clip exported ten times.

Then write one from scratch: describe the scene, the lines, and the pacing, and Riffkit generates an original video with your product and character, no source needed. Most people still start from a proven video, because that is what takes the guesswork out of the first three seconds. Creation mode is for the shots no existing video covers — a specific ad angle you need to test this week, a product moment nobody has filmed yet.

Default to riffing. Starting from a video that already won means the hook and the pacing are proven before you spend a thing, which is the whole reason your first three seconds stop being a guess. Create from scratch when no existing video covers what you need: a specific ad angle you want to test this week, a product moment nobody has filmed, or a scene you have already scripted. Creation also lets you set the exact length, anywhere from 4 to 45 seconds or picked by the model to fit your direction, which matters when you are filling a particular ad placement.

Specifics. When you create from scratch, the creative direction is the only source your script has, so a vague brief produces a vague video. Name the scene and who is in it, what the captions say, the lines, the pacing, the wardrobe, and how the product enters the shot. "Morning kitchen, low-calorie breakfast, opens on a yawn, caption reads 300 kcal in 5 minutes" beats "make a healthy food video" every time. When you riff instead, the direction is optional: the template already carries the structure, and your direction only steers it.

Yes, everything you make is yours to post, run as ads, and monetize, with no extra license or revenue share from us.