An AI model can generate a photograph or a video, but someone opening an app still has to decide what to ask for. Getting a useful result means choosing a tool and giving it clear instructions. Managing several subscriptions adds another complication.
For Sina Sinry, those problems became an opportunity to build on a business journey that began in e-commerce. In 2020, he reached his first $1 million in sales through Shopify. Running that business taught him how to sell online and attract buyers. It also gave him the capital to pursue something that fascinated him: mobile apps.
From selling online to building an AI product
As ChatGPT began attracting attention, Sinry imagined an app with chatbot assistants for everyday needs, including teachers and fitness coaches. He had no coding knowledge. The people he hired had little mobile app experience either.
He chose them for their character, willingness to learn and belief in his vision. “We learned and built at the same time,” he says.
Together, they researched competing products, studied how mobile platforms worked and learned how to acquire users. Sinry guided the team while learning an unfamiliar industry himself.
The product changed as he added image and video models. He saw two needs: access to multiple models through one subscription, and help writing the prompts that made those models useful.
“I try to think one step ahead of the market,” he says.
He paired sample images and videos with the prompts used to create them. The examples covered photoshoots, promotional clips, creator-style product videos and memes. People could choose a result they liked, examine its instructions and adapt them.
That approach also made prompting easier to understand. An image prompt can describe the subject, lighting, setting and visual style; a video prompt can add movement. Showing the instructions beside the result gave people a practical starting point.
Learning what an app actually costs
Building the product was one challenge. Making its economics work was another. “For one year, I was just losing money,” Sinry says.
He initially expected higher advertising spend to bring enough users to make the business work. Instead, the cost of acquiring a paying customer rose. App Store fees, advertising bills and AI generation costs all had to be covered. He considered quitting repeatedly.
His e-commerce experience had not prepared him for every part of a subscription business. He needed analytics to understand how long people stayed, what they spent and whether their payments covered acquisition and service costs. A growing sales figure could hide an expensive business.
Pricing needed attention too. Sinry says replacing monthly and annual subscription options with weekly and annual plans improved sales. He began treating the billing period as something to test alongside the product and advertising.
Finding an audience before paying for reach
His early attempt to delegate the search for successful TikTok formats had not worked. He took on the research himself, spending three months on the platform every night to understand its culture.
Eventually, one format connected. According to Sinry, it helped generate $7,000 in subscription sales in a day, followed by stronger days.
He developed a process around that search: running five to ten organic pages, posting daily and trying different formats. When a video passed 5,000 views, he would put paid advertising behind it.
Those initial views gave him a signal that a format was worth backing. Previously, he had gone straight to paid campaigns, spending money to discover which videos attracted attention. Organic posting let him do that initial testing before committing an advertising budget.
Flibbo, the app built through those experiments, has surpassed $2 million in cumulative sales, according to Sinry. His work in AI now extends beyond mobile. He also built a platform for the Middle East combining chatbot, image, video and music models, which he says exceeded three million users within six months.
He uses AI inside his businesses too. Sinry says he replaced his human assistant with an AI assistant connected through APIs to his analytics dashboards and advertising platforms. His mornings begin with a conversation about the reports.
The questions are ones he learned to ask during the difficult first year: what it costs to bring someone in, what they spend and whether enough remains to keep building.