Where would you cut?
The right moment doesn't wait.
Open source · Apache 2.0
Kairos turns any long video into vertical shorts ready to post. The AI proposes the best moments, then edits and captions each one and adds the sound. You choose, and nothing posts on its own.
interview-03.mp4 · 1:12:08 · 0 moments found
Why Kairos
The Greeks had two words for time. Chronos is time that passes. Kairos is the right moment. Your long video is chronos. Kairos finds the instant worth telling.
Inside a clip
Once a moment is picked, the AI works on the clip itself. It fixes the transcript, chooses the words to highlight and places every sound on a real spoken word. It all lands on the timeline, where you can change anything.
The AI fixes misheard words and names, highlights up to two words per phrase and translates each phrase. Every word stays tied to the instant it is spoken.
Whooshes on the cuts, a pop on the key word, an impact with a punch-in on the line that lands. The music drops under the voice and the clip is normalized to −14 LUFS.
How it works
A file, or a YouTube, Twitch or Kick link. You state its rights before processing.
It transcribes the video and analyses the voice, loudness and scene cuts. Every proposal comes with a score and a reason.
Tick the moments and pick the model, framing, caption style and languages.
Hook, your own commentary, animated captions, effects, color, sound and metadata for each platform.
Fine-tune in the timeline editor and approve in the queue. Nothing posts on its own.
Features
Synced to every word, in 8 styles, with the key words highlighted.
Captions and metadata in each language, translated naturally.
10 color looks, zooms, shakes, flashes, animated titles, sound effects and music.
Title, description and hashtags for YouTube Shorts, TikTok and Instagram Reels.
NVIDIA, AMD or Intel when available; otherwise, the CPU.
Live preview, five tracks, undo and autosave. What you see is what renders.
Claude, OpenAI, Gemini, OpenRouter or a local model, with your own API key. No servers of ours, no telemetry.
Three downloads, one app
Download the EXE and follow the setup. No need to install .NET, Python or FFmpeg. First-run setup asks for a folder, your AI provider and your API key.
Extract the whole ZIP to a writable folder and open KairosStudio.exe. Keep everything together when moving this copy.
Extract the bundle and run the command below. Docker builds the app, then opens the web panel at localhost:8000. Docker and internet are required.
$ docker compose -f compose.yaml up -d --build
Windows builds are currently unsigned and may show an unknown publisher warning. Downloads are hosted on GitHub. SHA-256 checksums · Release notes
FAQ
Kairos is free and open source. You pay your AI provider directly with your own API key, or nothing at all with a local model. The app shows how much you have spent and how much of your budget is left.
Claude by default, or OpenAI, Google Gemini, OpenRouter, Mistral, DeepSeek, Groq, or a local model through Ollama, LM Studio or vLLM. Outside Claude, you pick the model for each tier in Settings. The prompts were written and tested with Claude, so results can vary with other models.
No. Videos, clips and transcripts stay on your machine. Only the transcript text and a few frames go to the AI provider you choose, to decide the edit. With a local model, nothing leaves your computer.
No. With an NVIDIA, AMD or Intel GPU rendering is faster; without one, Kairos uses the CPU.
No. Every clip goes through an approval queue and a person decides. Kairos exports the video, captions and metadata ready to upload.
Your own, or ones you have permission or an open license for. The app asks for the rights status and evidence before processing.
Yes, with the Docker version. The desktop app is for Windows 11.
Open source, with your own API key, on your own machine.