Investigation · 6 September 2026 · 5 min read
Every list of Lebanese AI startups in circulation names the same ten companies, and they all cite the same source. That source is a coding bootcamp's marketing blog, and its funding figures are wrong by a factor of nine.
Search for Lebanese AI startups and you will find the same ten names. NAR. eFlow.ai. NuWatt. Mindmosh AI. Silni. Katana. Zaker. AirGuard One. LUXEED Robotics. Neumann. They appear in roundups, in investor decks, in LinkedIn posts about the resilience of the Lebanese tech scene, and in at least one AI generated research report we were sent this month.
Follow the citations back and they converge on a single page: a blog post published by an American coding bootcamp as content marketing for its own courses. It is the origin of almost every Lebanese AI list currently in circulation.
That would be tolerable if the numbers held. They do not.
The post prices each company in Lebanese pounds and then helpfully converts to dollars. LUXEED Robotics is listed at 15 billion Lebanese pounds, which the post calls roughly 1.5 million dollars.
The official exchange rate is 89,500 pounds to the dollar. Fifteen billion pounds is about 168,000 dollars.
The conversion is wrong by a factor of nine, and the same factor runs through the whole list. NAR at 20 billion pounds becomes 2 million dollars instead of about 223,000. eFlow.ai at 7.5 billion becomes 750,000 instead of about 84,000. Katana at 2.5 billion becomes 250,000 instead of about 28,000. Every figure is inflated the same way, which tells you the numbers were not checked individually. They were generated.
No other outlet has published these figures. There is no filing, no press release, no named investor and no funding announcement behind any of them. They exist in one place and get copied outward.
The companies themselves may be entirely real. Several almost certainly are. That is not the point. The point is that a number nobody can trace has been laundered into a fact by repetition, and it is now the primary quantitative claim anybody makes about Lebanese AI.
The damage is not that a bootcamp got some maths wrong. It is that Lebanon has, at this moment, an actual AI story worth telling, and it is being buried under an invented one.
Here is what is verifiable.
HAQQ Legal AI raised 3 million dollars in February 2026, led by Sowlutions Ventures. It was reported by Wamda and by L'Orient Today, from Beirut. Founded in 2020 by Antoine Kanaan and Abbas Kabalan, it runs a legal AI engine that produces jurisdiction specific output, with practice management and payments built on top, and it reports more than 7,000 clients including bar associations and courts. This is the largest confirmed AI round in Lebanon and it is missing from most of the lists that name Silni.
World Monitor was built in a single day by Elie Habib, co-founder of Anghami, the only Lebanese founded company on NASDAQ. It maps global events in real time from more than 450 verified sources, with a rules based severity engine, pattern detection for escalation, and a language model doing final classification. It is open source. Within weeks it had more than four million visitors from 174 countries. It appears on the bootcamp list, credited correctly to Habib, and then appears nowhere else with the traffic figures that make it the most used piece of software any Lebanese engineer has shipped this year.
Bounty raised 1 million dollars from Andreessen Horowitz. Rayan Dabbagh left Beirut at seventeen, went from the Grand Lycee Franco Libanais to Georgia Tech to AWS, and now runs a marketplace in New York where AI agents claim and complete paid work, with the money held in escrow until a result is verified. Not one of the Lebanese AI lists includes it, because it is not in Lebanon.
Those three facts are all checkable in under ten minutes. The nine times error survived because nobody spent the ten minutes.
The reverse error is just as common, and it flatters rather than deflates.
Medicus AI is described almost everywhere as Lebanon based. The company's own about page says it was founded in Dubai in 2015 and moved its headquarters to Vienna in 2016. Beirut is one of five offices. The Lebanese connection is genuine and it is a good one, since co-founder Nadine Nehme did her doctorate under a Nobel laureate, but it runs through a person, not through a headquarters.
Karim Atiyeh, born and raised in Lebanon, co-founded Ramp and is now its Co-CEO. Most Lebanese coverage still calls him CTO. It has been out of date for a while.
These are not pedantic distinctions. A country that can name three companies headquartered inside its borders is in a different position from a country whose best people are running companies somewhere else, and the honest version of the Lebanese AI story is that both things are true at once. Collapsing them into one number produces a figure that is useless to a founder, to an investor and to anybody deciding whether to stay.
We have published a companies page under the same rule that governs our calendar. An entry needs a named founder or organisation, a page we opened ourselves on a stated date, and a printed line saying exactly where the Lebanon connection comes from and what its evidence is. Companies based in Lebanon and companies run by Lebanese founders abroad are separated, permanently, because they are different claims.
Nine entries made the page. Twenty one were opened and left off, and the reasons are published beside them rather than kept private.
Where our evidence is thin, the card says the evidence is thin. Lexyom, one of the earliest Lebanese legal AI attempts, is listed with a note that we have not independently confirmed it is still trading. That note is not a weakness in the page. It is the only thing that makes the rest of it worth believing.
There is money arriving too, which is the part that makes accuracy suddenly expensive. B&Y Venture Partners has launched LAIN, a 5 million dollar fund aimed at Lebanese startups in AI, fintech and SaaS. Funds do diligence. Sooner or later somebody is going to open one of these lists, try to verify a number, and find the bootcamp blog at the bottom of it.
Better that it is us, and that we said so first.
Sources
Written for Club Cedro. If something here is wrong, say so and it gets corrected with the correction left visible.
More writing