Play. Validate. Earn.
That's the whole loop.
It's the tagline and the business model in one: players have fun and get paid; the crowd's agreement turns play into trustworthy data.
Play.
Pick a game - anything from diacritics to dialects to dog breeds. Sessions are 30 seconds to a few minutes. It just feels like trivia.
- No surveys
- No queues
- Any topic, any game
Validate.
Behind the scenes, the same item is shown to several players across different games. Where the crowd agrees, the label is trusted. Where it splits, it goes to a tie-breaker.
- Many eyes per item
- Agreement = quality
- Gold checks catch bad actors
Earn.
Correct, agreed-upon answers mint coins. Cash them out through the methods available in your country, shown in the app.
- Rate shown in the app
- Methods vary by country
- Cash out for real
Tens of thousands of conversations.
Nobody has read them.
Your dashboards count your agent's conversations. They don't tell you whether the answers were right. Our reviewers read them and show you what to fix.
Every game is
secretly collecting data.
A few of the games we've built, playing live. Swipe through them and watch every round turn into a labeled data row.
Tashkeel
Read the sentence, tap the right diacritic on each letter.
Each round becomesعلمعَلِمَ
Who are you?
Frequently asked questions
What is Avenzoar?
Avenzoar is a gamified data annotation platform. Companies and researchers bring us a data need, such as text, speech, dialects, images or answer ratings, and we turn it into short mobile games. Real people play, several players cross-check every item, and their answers become human-verified AI training data.
How do games turn into AI training data?
Every item is answered by several independent players, often across different games. Hidden attention checks measure how reliable each player is, a label is accepted only at trust-weighted consensus, and disputed items are served again and reviewed by an expert. Each label comes with the agreement behind it.
What kinds of data can you label?
Anything a person can judge: text and sentiment classification, named entities, semantic similarity, fact checks, ratings of LLM answers, content moderation, dialect identification, Arabic diacritics (tashkeel), OCR and handwriting checks, and speech transcripts. We put your data into one of our existing games or design a new one around your task.
Is Avenzoar only for Arabic data?
No. The games run in English and Arabic and work for any topic. Arabic is our focus: dialects, diacritics, handwriting and speech, where labeled data is scarcest.
How can I play?
Download Avenzoar for Android on Google Play. The iPhone app is not available yet. You can also try a data-labeling game in your browser right now, without signing up.
Where is Avenzoar based?
Our headquarters are in Amman, Jordan, with operations in Doha, Qatar. Avenzoar is incubated by Qatar Science & Technology Park (QSTP) and Orange Corners.