Smart triage
Every message is classified instantly: news, reply, or chatter — reactions are excluded automatically.
An integrated AI system
Six technologies working together behind the scenes — from the moment a message arrives until the story is drafted.
Built around the journalist — not instead of them
The AI prepares, filters and suggests; the editorial judgement stays fully human.
01
02
Confidence per report
A numeric score + the gap list + corroborating reports — never an automated "confirmed".
03
Location extraction
Governorate, district and street extracted from the report text by AI.
04
Image & video analysis
Computer vision that understands visuals and reads text inside photos.
05
Headline drafting
An official headline, a social headline and a summary — ready to edit in one click.
06
Disciplined intake agent
Gathers details and evidence from the source — three questions max, then a journalist handoff.
From noise to signal
Every message flows through a connected pipeline of AI agents — only what matters reaches the journalist.
01Incoming messageFacebook · Instagram · WhatsApp
02Triage agentNews or chatter?
03Attribution agentConfidence + gaps
04Intake agentEvidence from the source
05Journalist's callThe human decision
Every message classified in seconds
Story replies and reactions are marked "not important" automatically — without consuming any AI resources.
News
"Breaking: a truck overturned on the agricultural road — traffic has stopped"
A text report entering the AI pipeline instantly: priority, location, confidence.News + video
"Footage from the fire scene before firefighters arrived"
The video is archived locally and analysed with computer vision.News + photo
"A photo of a sudden sinkhole in a side street"
The image is analysed by AI — including reading any text inside it.Not important
"Hello 👍❤️" — a reply to a published story
Greetings, reactions and story replies are excluded with zero processing.
Why Nour?
A journalist's time is for decisions —
A journalist's time is for decisions —
not sifting
In real newsrooms, about 93% of inbound messages aren't news. Nour absorbs that entire volume with AI, and puts a short, classified, scored list in front of the journalist.
- Urgent reports arrive with a red badge and an instant alert
- Media is archived locally so it never expires
- Smart filters search the whole dataset, not the latest message
- A human-review queue for every case that needs a decision
By the numbers
AI built on real data
Nour wasn't built in an isolated lab — it was built on real data from a working newsroom.
14K+real messages the system was built on
93%of inbound traffic filtered automatically
3questions max before a journalist handoff
6AI technologies working together
One platform for the whole editorial team
The inbox journalist
Opens an AI-sorted inbox instead of a sea of messages — urgent reports arrive with instant alerts.
The verification editor
Receives each report with its confidence score, gaps and corroboration — starting halfway there.
The managing editor
Tracks analytics, source performance and the review queue — and tunes the agents with clear switches.
Articles about Nour
The full story: why we're building it, how its technology works, who its agents are, and where it's going.
Why newsrooms need Nour
Manual sifting and its cost to the scoop.
Read the article →
From message to story
A message's journey through the AI pipeline.
Read the article →
The three AI agents
Roles, limits and design philosophy.
Read the article →
Technology & roadmap
Gemini, the platforms, and what's next.
Read the article →Questions?
Does the AI decide a story is "confirmed"?
No — by design. Nour gives a confidence score, the missing details and the number of corroborating reports; the final editorial decision always belongs to the journalist. Our framing is "attribution & confidence", not "automated verification".
What happens to reactions and story replies?
They're marked "not important" automatically and instantly — even with long text or media — because they're engagement with the page's content, not reports. This saves journalists' time and AI running costs together.
Can the intake agent promise publication or judge a story?
Structurally impossible: its messages are fixed templates, not free-form LLM text, with a three-question cap and a mandatory journalist handoff. The switch for actually transmitting replies is separate — and OFF by default.
Which platforms are supported?
Facebook Messenger, Instagram and WhatsApp via Meta's official APIs. Media is archived locally to survive the platforms' expiring URLs.
What powers Nour's AI?
Google Gemini — one unified provider for every capability: triage, classification, location extraction, headline & summary drafting, and image/video analysis. All of it works in Arabic.
Are there user roles?
Yes: admin, editor, user and viewer — with secure sign-in and optional two-factor authentication.
Ready to reach the real story faster?
Enter the platform and try the AI capabilities yourself — triage, attribution and intake, running on your real data.
Enter the platform