Five ways to make money from AI in Sudan, and why none of them pay tonight

You can make money from AI in Sudan. You cannot do it tonight. Aziz says the second part first: anyone who tells you the money arrives immediately is either lying to you or selling you something. He refuses the opposite pose too. One day he thinks the technology will finish us all, the next he thinks nothing better has happened to us. He does not know, and the people writing it do not know either.

What he is sure of is narrower and more useful. The tools are available today and there is a benefit in them today, so deal with today and see what tomorrow brings. He planned twenty routes for the episode and cut them to five. None needs a course, because he never took one. He used the tool to teach him the tool, which is the only training anyone has had. An edit that used to sit with a designer for a day and an editor for two or three now comes back half an hour after he writes the prompt.

Before any of the five routes comes the question that stops most people: how the money actually reaches you. Sudanese banks do not have the international relationships that let someone receive payment from outside. He adds, in the same breath, that he may be wrong about this, and if he is, ignore him and carry on.

The workarounds are the ones people already use. Almost everyone has a relative abroad: the money lands in their account and they send it on. Every city with Sudanese in it has someone running transfers home. Failing that, earn first, get the security approval, travel to Egypt, open an account there, and go out every few months or find someone who can act for you.

He met a man in Sudan, early twenties, who opened a YouTube channel around 2013. He never appeared on camera. He collected news and published it, reached a hundred thousand viewers, earned thousands of dollars, and solved the banking problem by opening an account in Egypt. That man had no AI. You do. One exception: if your customers are inside Sudan, none of this applies.

The first route is content, and it is where he would start, because it already works and you can watch people doing it. The Impact Lab publishes ten reels a week, he publishes three or four of his own, and an episode goes out every two weeks. That costs him about an hour a week, or rather four or five hours in one day covering the whole month. Two years ago the same output would have taken days of a designer's time and days of an editor's. Quantity matters as much as quality when you are building an audience, and once the audience exists you have options: sponsorship, your own product, paid promotion.

Do not ask him what to post, ask the tool. Tell it you are in Madani and want an audience in Gezira, and ask what pulls people there. Then argue with the answers and run tests. The audience analysis that used to take his agency weeks of surveys is now a question.

The second route exists because of a weakness. The models are poor at Sudanese Arabic, and he knows it because his own subtitles come out wrong and he cannot read Arabic to fix them. Organisations and YouTubers with large Sudanese audiences publish in formal Arabic, in Egyptian dialect or in English. Go to them and name the problem they may not know they have. Let the model produce a first pass, correct it into Sudanese, feed your corrections back, and after fifty or a hundred rounds your model understands the dialect better than theirs. A large Sudanese client once had an agency in Egypt push to run the same campaign, in Egyptian dialect, inside Sudan. He told them plainly that this is the easiest way to make Sudanese people dislike you.

The third route is the least glamorous and the one he would do immediately. Go to businesses in Sudan: every shop, restaurant, bakery, clinic and dentist, and the woman making kofta at home to sell to her neighbours. Most have a WhatsApp business account, almost none have a catalogue in it, and most are not on Google Maps. Photograph the products on a phone, hand them to the tool, ask it to upscale, cut the background and write the listing. If you do not know how, ask the model to explain it step by step as though you were fifteen with a new phone. Then do it free five times, use those five as your examples, message twenty businesses you already know, and charge half up front and half on delivery.

The fourth route goes to the AI companies. They want their models to work in languages they cannot reach, and Sudan is a high threshold for them, close to impossible from where they sit. You are Sudanese, which is the entire qualification. Gather people, record real Sudanese speech, voice only, no faces, grandmothers included, and hand over a transcript in English they can train on. His reference is India, where women are paid daily to carry a phone that records how they speak. The market is real and is already being served somewhere else.

The fifth is written work and research, and it is the slowest. Most of what the European Union, the United Nations and the large organisations publish about Sudan was not written by a Sudanese person, and those bodies already move money into the country. Approach them and point out that everything written about Sudan is written from outside it. If they find someone they trust, who is consistent and delivers when they said they would, there is effectively no ceiling on what they will pay.

No one is going to pay you tonight, so be ready to work and work and work until you see a single coin.

If you keep going, income comes. If it never comes, you will have learned a great deal about yourself, about the tool and about the market. And if you are sitting at home without a job, he asks the obvious question: what exactly are you losing.

One warning before anyone starts on content. The algorithms favour the shocking, and he has watched where that leads, one video at a time, until people are doing things they never set out to do. His own filter is simple: he does not want his mother to be ashamed of anything he has made.