Kawuri can now make and see: images, video, voice and media analysis in Indigen World 0.1.20

Kawuri has always been the guide you could ask about Kassena culture and the Kasem language. In Indigen World 0.1.20 it learns to make and to see. You can create an image or a short video from a description, speak a message instead of typing it, and ask Kawuri what a photograph, a clip or a recording shows. It does all of this carefully: it keeps what it can see apart from what it is guessing, and it never passes a guess off as fact about a people or their language.

What is new at a glance

Open Kawuri from the Learn tab and you will see the tools along the top of the home screen. Tools that used to say "Coming soon" now work:

• Create image: describe a picture and Kawuri makes it, in the shape you choose.
• Create video: a short clip of four, six or eight seconds, for approved creators and members on the Indigen Creator plan.
• Voice input: tap the microphone and speak your message in English. Kawuri turns it into text you can edit before sending.
• Analyse media: attach a photo, video or audio file, or take a photo, and ask what it shows.
• Recent: your latest creations now sit on the home screen, with their real status.

If a tool is not available to you, Kawuri says why before you start. You might need to sign in, or your plan might not include it. You will never fill in a whole form only to find out at the last button that nothing happens.

Create an image

Tap Create image. Describe what you want to see, then choose a shape: square, portrait, landscape, tall or wide. Kawuri only offers shapes the image model can make.

1. Describe it. Be concrete about the scene, the light and the mood. "An illustrated storybook scene: a woven basket and clay pots in a courtyard with painted geometric walls, morning light, no people" works better than "a village".
2. Pick a shape. Portrait and tall suit a phone screen and a reel cover; landscape and wide suit a website banner.
3. Add a reference picture, if you like. Kawuri uses it as a guide for colour, composition or style. Pictures of real, identifiable people are not accepted.
4. Create. An image usually arrives in under a minute.

When it is ready, you can:

• Save it to your device, or share it: a shared image goes with a note saying it was made with AI.
• Use it as a reel cover. Kawuri starts a reel draft with the picture as its cover, and you add your video.
• Use it in a contribution, as the artwork for a song or recording. The contribution form's usual questions about sources and rights still apply, and reviewers see that the artwork was made with Kawuri.
• Regenerate it for a fresh take, edit the prompt and try again, or delete it.

Create a short video

Create video makes a clip of four, six or eight seconds, in portrait (9:16) for phones and reels or landscape (16:9) for wider screens. Landscape can be made at 1080p; portrait is made at 720p. You can describe what to leave out, such as text on screen or watermarks, and start the video from a reference picture.

A video costs far more to make than an image, so three things are different:

• It is for approved creators and the Indigen Creator plan. Everyone else sees why the tool is not available to them rather than a form they cannot use.
• It asks first. Before anything is made, Kawuri asks Use a video generation? and tells you it uses one of your limited AI videos for the day. Nothing is started until you confirm.
• It takes a few minutes, and you do not have to wait. Close Kawuri, lock your phone, go and do something else. The video keeps being made on Indigen World's servers, and you get a notification when it is ready or if it could not be made.

Why the videos have no sound

Kawuri's videos are made silent, on purpose. The video model would invent speech and song, and it cannot speak Kasem. We will not put a made-up voice in a language that is not Kasem over footage meant to show Kassena life. If a video needs a voice, it should be a real one. Add your own recording when you use the video in a reel.

Speak your message, in English

Tap the microphone below Kawuri's message box. The first time, Kawuri explains why it needs your microphone before your phone asks for permission. Then record, pause if you need to, listen back, and tap Use as text.

Your words appear in the message box as text. Nothing is sent until you read it, fix anything that is wrong, and press send yourself. If a word was unclear, Kawuri marks it [unclear] instead of guessing. The recording is uploaded privately, turned into text, and deleted straight away. A voice message can be up to two minutes long.

We know this is the wrong way round for a project about Kasem. Speech recognition for Kasem needs recordings that speakers have checked and agreed to share, and that collection is still being built through Contribute. English comes first because it can be done well today. Kasem will come when it can be done properly.

Ask Kawuri what is in a photo, clip or recording

Choose Analyse media, then attach a picture, a video (up to five minutes) or an audio recording (up to ten minutes) with +, or take a photo with the camera button. Pick what you would like Kawuri to do:

Add a question if you have one, and send. Then keep asking about the same file in the same conversation: "Is it sharp enough for a reel cover?", "What text is on the sign?"

Seen, possibly, and unknown

Media about culture is easy to get confidently wrong. A picture of a painted wall can be described with certainty; who painted it, where, and what the patterns mean usually cannot be read from the picture alone. So every analysis keeps its answer in separate, labelled parts:

• Seen directly: only what is actually visible or audible.
• Possible context · not verified: likely interpretations, each worded as a possibility.
• Suggested languages · not verified: a guess, never a confirmation.
• What cannot be known from this: what the media does not show, and why.

Whenever an analysis touches on culture or language, it ends with a reminder that it needs community verification. Kawuri will not state as fact who a person is, their ethnic group, which language is being spoken, what ritual or sacred object is shown, where something happened, or what a symbol means, unless the media itself shows it plainly. When it does not know, it says so. And it does not identify people.

Your creations, in one place

The Recent section on Kawuri's home shows your two newest creations with their real status: Image · Ready, Video · Generating, and so on. Tap See all for the full library, filtered by All, Images, Videos or Analysis.

Each creation's menu only offers what makes sense for it right now. A video still being made can be cancelled. A finished image can be saved, shared, used as a reel cover or in a contribution, regenerated or deleted. An analysis can take a follow-up question. Deleting a creation removes its file and its details from your account.

Creations are private. Only you can see them until you choose to save, share or use one somewhere else.

Safety and respect, built in

Before Kawuri makes any image or video, Indigen World checks the request, and any reference picture, against our own rules. We added this check after testing showed the AI provider's filters alone were not strict enough for us. Kawuri declines to make:

• real, identifiable people, including public figures, or a photo of a real person to be changed or animated
• anyone who appears to be a child
• sexual content, graphic violence or injuries
• hate, harassment or demeaning stereotypes of any people
• realistic scenes of news, crimes, disasters or politics that could be mistaken for real
• sacred rites, shrines, masks or ceremonies presented as if they were real documentation of a community

When a request is declined, Kawuri says why in plain words, so you can describe it differently. Illustration, invented characters, landscapes, crafts, food, music, markets, farming, festivals and folktales are all welcome.

Everything Kawuri makes is marked as AI-generated in the app. When you share a creation, or use it in a reel or a contribution, Kawuri adds that disclosure for you. Please keep it there when you post it elsewhere.

Translations still come from the dictionary

None of this changes how Kawuri handles Kasem. When you ask how to say something, Kawuri looks in the published Indigen World dictionary first, quotes the entry exactly, and says so when the dictionary does not have the word. If it adds anything beyond the dictionary entry, such as usage advice or background, that sentence now starts with Not verified:, and Kawuri invites you to confirm or correct the word with the community.

Allowances

• Each image, voice message and analysis question counts as one message against your plan's daily Kawuri allowance, the same allowance your chat uses.
• Videos come from a separate daily AI video allowance, shared between Kawuri and TribeStudio.
• Tapping twice, or retrying after a dropped connection, never makes a second image or video. Kawuri recognises it is the same request.

You can see your plan in My Space → Membership. Allowances reset within 24 hours.

How it works, briefly

Kawuri's new tools run on Google's Gemini and Veo models on Vertex AI. They are reached only through Indigen World's own servers, never directly from your phone. The app holds no AI keys or passwords. Your phone uploads media to a private folder only you can read. The server checks your account, checks the request, makes the creation and saves it privately.

A video is tracked on the server from start to finish, so it survives you closing the app, and even our servers restarting. A task that takes too long is stopped and reported as failed; nothing is ever left "generating" forever.

Not here yet

• Kasem voice input. English only, for the reasons above.
• Editing an image you already made. You can edit the prompt and create a new one.
• Sound in videos, deliberately. Add your own recording in a reel.
• Kawuri speaking its answers aloud, and answers that appear word by word as they are written.

Also in this update

• Confirmations in Kawuri (creating a video, cancelling, deleting) use the same card as the rest of the app.
• A generated video or image can be handed straight to the contribution form or a new reel draft, with its AI disclosure already filled in.

Try Kawuri's new tools, and tell us what you make, and what it gets wrong, in the Community tab. If an analysis gets something about Kassena life wrong, reporting it helps everyone who uses Kawuri after you.