Osegi
AI·2 min read

Qalu, an AI app that reads Arabic dialects and spots PR crises

Qalu reads text, images and video in Arabic dialects, so it can tell when a angry "thanks a lot" is actually a complaint. Here is what the app does.

A dashboard showing social media posts in Arabic with sentiment labels, charts and crisis alerts on a laptop screen.
Image: اليوم السابع
SUMODialoediviewergiftdetect

Picture an angry customer writing «تسلم إيدكم», literally "bless your hands". In Egyptian Arabic that can be praise or a bitter jab. Many sentiment tools score it as a compliment. Qalu, a new AI app from the Egyptian firm Infotimes, is built to catch exactly that kind of mistake.

What is Qalu?

Qalu is an app and platform that analyses what people say about a brand across the web. Keyword trackers miss a lot: a photo of a damaged product with no company name, a video complaining about bad service, a sarcastic comment. According to Infotimes, Qalu reads text inside images and describes their content, transcribes and summarises speech in videos, then combines all of it with posts and comments.

It supports Modern Standard Arabic, Egyptian, Gulf, Levantine, Iraqi and Maghrebi dialects, plus English, and it translates content posted in other languages.

How does it spot a crisis?

Founder and CEO Amr El-Iraqi says the idea came from years of media monitoring work. The hard part, he says, is not counting words but understanding what the audience means. Qalu tries to separate a direct accusation from a news story that merely reports one, and to work out who is being talked about. It shows the phrases behind each label so users can check the result.

The app pulls content from news sites, Facebook, X, Instagram, TikTok, YouTube, LinkedIn, Threads, Reddit and Bluesky, plus public Telegram channels, forums, blogs, podcasts, and reviews on maps and app stores.

El-Iraqi says some organisations only learn of a crisis once complaints reach the news. Qalu looks for earlier signs, such as a sudden jump in negative posts or a video spreading unusually fast. Alerts go out through the platform, email and phone, and urgent ones escalate if nobody opens them. A network analysis shows who posted a topic first, who reshared it and which accounts widened its reach.

Users can also question the data in plain language, for example "what did customers complain about most this week?" The answers come with figures, charts and links to the original material.

Infotimes was founded in 2014 and works in data journalism, big data processing, training and AI tools. It won the Global Editors Network award in 2018.

Apps worth a look

SUMO: the number puzzle

Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.

Tags:qaluinfotimesarabic aisocial listeningsentiment analysismedia monitoring

More from this day

See all →

More in: AI

See all →