You're on a street in Tokio looking at a sign you don't understand. Or in Pamplona facing a pintxos menu full of words a tourist can't decode. You pull out your phone, open the KAIXO menu, tap Photo Translator, point — and two seconds later you're reading what it says in your language.

What the Photo Translator is

The Photo Translator is a new feature inside KAIXO that turns any text in the real world into text you can understand. You point the camera, snap, and the app extracts the text in the image and translates it into the language set in your profile — or any of the other 15 KAIXO supports if you change it in the dropdown.

It's not generic OCR or Google Lens bolted on. It's Claude Vision (Anthropic's multimodal AI) reading and translating in a single pass, with the translation appearing letter by letter as it's generated — no dead time.

How to use it, step by step

  1. Open the KAIXO main menu (three dots top-right). There you'll find "Photo Translator" with its own icon.
  2. Pick the target language in the dropdown above the camera. By default it's yours, but you can change it if you want to show the translation to someone else.
  3. Point and shoot. Like a normal camera. If your phone has an ultra-wide, there's a 0.5× zoom chip to cover bigger signs.
  4. Read the result. The detected original text appears, the translation below highlighted in orange, and underneath buttons to listen (TTS), copy to clipboard or take another photo.
  5. Tap Listen if you want to hear how it sounds in the target language — useful for repeating the exact name of a dish to the waiter.

Tip: for full menus it's better to step closer and shoot by sections (starters / mains / desserts). A single photo of the whole menu loses resolution, and the AI can skip small lines. 2–3 clean shots beat one blurry panorama.

Real cases where the Photo Translator saves you

This isn't an abstract tool — these are situations where you actually feel the difference:

  • A restaurant menu in another language. You arrive at a place, they hand you a menu in Japanese, Italian, French. You snap a dish, you read "potato omelette with caramelized onion and piquillo peppers" and you know what to order. If you're allergic to something, the original stays visible for the waiter to confirm.
  • Notices and opening hours. That closed door with a paper stuck on it in German, where you don't know if it says "opens at 5 pm" or "closed for holiday until September". Photo, one second wait, answer.
  • Street signs and directions. Subway exits in Seoul, park rules in Berlin, vending-machine instructions in Istanbul. Things that aren't important enough to open Google Translate and type, but that you skip and end up paying for.
  • Helping a tourist who comes to your business. They bring you a piece of paper with a question or an address scribbled by the hotel. You switch the target language to yours, snap the paper, understand it, and answer with KAIXO Direct or in writing.
  • Short documents in another language. A receipt, a parking ticket, a safety notice at a construction site. Targeted text where you want precision without making it a whole project.
  • Going somewhere with an older relative who no longer handles foreign languages. Your father reads better in Spanish than in French and you're traveling in Bordeaux. You hand him the phone, photo at the menu, done. Older people appreciate being able to read in their own language without asking for favors.
Person aiming a phone camera at a sign in a market
The motion: focus calmly, shoot cleanly, read in your language. Photo: Unsplash.

Differences from Google Translate Lens

The logical question: doesn't Google do this? Yes, it does something similar — but with practical differences that matter in real life:

  • Literary translation, not literal. Google Lens overlays word-by-word on top of the original sign. KAIXO Photo Translator returns a clean block, translated the way a human would write it — without those robotic "pintxo de tortilla → small pintxo of omelette". For local dishes with no equivalent, it keeps the original name and adds an explanation in parentheses.
  • Designed for dishes and signs, not documents. If you're in a tourist area, 80% of what you'll want to translate are menus, directions and notices. KAIXO is tuned for that use case, not for scanning contracts.
  • Integrated with the rest of KAIXO. After reading the sign, if you want to ask the waiter something, you open KAIXO Direct and say it by voice. If you want to share what you're about to order with a friend in another country, photo + send to chat. All in the same app, no jumping between five.
  • Privacy by design. The photo is sent to the server only to be processed at that moment; it isn't stored, doesn't train models, doesn't appear in any history.

We're not saying Lens is bad. It's a massive piece of free engineering. What we're saying is that for the case "I'm in front of a menu or a sign and I want to understand it well", KAIXO Photo Translator is built specifically for that, and you can tell.

Languages, connection and privacy

15 languages: Spanish, English, French, German, Portuguese, Italian, Basque, Chinese, Japanese, Arabic, Hindi, Russian, Korean, Turkish and Indonesian. Works in any direction — from the detected original to whichever you want.

Needs internet. Image recognition + translation happens in the cloud. Wi-Fi or mobile data both work. Without a connection it can't operate — an honest limitation of the model, not a bug.

Live streaming. Instead of waiting 5–10 seconds for everything to finish, the translated text appears character by character as Claude generates it. Sense of immediacy without tricks.

Privacy: the image is sent encrypted, processed, discarded. Not stored on the server, not used to train. Full policy.

When NOT to use the Photo Translator

Honestly: there are cases where other tools are better:

  • Long documents (contracts, books). For long text, better to copy to clipboard and use a long-block translator. Photo Translator is built for short visual fragments.
  • Unclear handwriting. The AI may confuse letters if handwriting is very personal. Chalkboard menus tend to work; recipes scribbled by a doctor — not so much.
  • No internet. If you're going into a dead zone (mountains, plane, some metro lines), download offline phrases in another app. Photo Translator needs a connection.

Get started

If you already have KAIXO installed, you'll find it in the main menu: three dots top-right → Photo Translator. The first time you open the camera, the system will ask for permission — grant it and you're set.

If you're coming to Pamplona for San Fermín, traveling to Tokio or Bangkok, serving tourists at your job, or simply want to understand the world around you better without having to learn 15 languages — download the app and try it.

FAQ

Does it work with any written language? Yes, the 15 KAIXO supports. If the original is in one of them and your target is too, you're good.

How long does it take? Typically 2 to 5 seconds. You'll see the text appear in streaming.

Can I send the translated photo to chat? In the current version we moved Photo Translator to the main menu so it's always at hand. If you want to share the result, tap "Copy" and paste it into the chat you want. We'll add a direct button in upcoming versions.

Does it cost money? There's a generous free allowance for personal use. If you use it intensively (business), paid plans offer unlimited volume.

Does it work in landscape? Yes. The buttons rearrange automatically so they're comfortable to tap when holding the phone sideways.

What about the ultra-wide camera? If your phone has multiple rear cameras (most from 2020 on), a 0.5× chip appears to widen the angle. If there's only one camera, the chip doesn't appear — it still works at 1×.