Honestly, it's time to move past the endless tweaking of Midjourney parameters. Last month, I landed my first custom AI couple photo order for just 99 yuan. The clients were a long-distance couple looking to post something special on social media for their anniversary. From receiving the images to getting paid? Under 20 minutes.
Their request was straightforward.
Many people find this whole thing mysterious, feeling the need to fully understand every Stable Diffusion model. Truth be told, it’s not necessary. Users don’t care whether you generated the images with SD 1.5 or XL—they just want to know if the image can earn them likes on their posts or help fill the gap of not being able to travel together.
This is an emotional business, not a tech competition.
I closed my first loop using tools that were embarrassingly simple—just some readily available AI art platforms, sprinkled with a bit of my own aesthetic tweaking. The key? Find that group of people who are eager to share photos but have limited budgets or find photography a hassle.
Xiaohongshu (Little Red Book) is a goldmine.
I searched for “couple avatars niche,” and the requests were pouring in. I did something super simple: I generated several sets of couple avatars in different styles, created a comparison image, and posted it with the title “The cool couple avatars you wanted, AI can do it in five minutes.”
And just like that, the traffic came rolling in.
I received DMs asking if I could do custom work, so I quoted 9.9 yuan for a trial edit. This not only validated the demand but also helped me pinpoint the lowest price users were willing to pay. It’s way more practical than spending a week discussing “whether there’s a market” in a group.
But there’s a huge misconception here.
You might think users want something that looks just like them, but what they really want is the “vibe.” I’ve had people ask if a real-life photo was AI-generated simply because the lighting gave it a cinematic feel. So instead of focusing on every detail, it’s more effective to study trending filters and compositions that capture the right emotions.
That’s how you capitalize on the information gap.
I later packaged my services into three tiers: 9.9 for quick images (simple backgrounds), 39.9 for refined edits with captions, and 99 for an anniversary package (three different scenes). Surprisingly, the most popular option turned out to be the priciest one because users felt it was “worth it”—they avoided all the hassle of planning, makeup, and venue booking, and instead received a set of “stunning” images ready for social media.
Technology solves the question of “can it be done,” while insights determine “is it worth buying.”
Of course, this isn’t just about sitting back and letting the money roll in.
The biggest challenge is communication costs. You’d never guess the kinds of requests users come up with: “Can you edit my monolid into double eyelids?” or “Can you change the background to the café where we first met?” I lacked experience at first and nearly got overwhelmed by edit requests.
I wised up and created a standardized process and case library. Now, I send examples upfront: “Hi there, does this style work for you?” This way, I manage expectations from the start and eliminate about 80% of the back-and-forth.
Another pitfall is copyright. If users take the AI-generated images for commercial use and run into trouble, who’s responsible? So I made it crystal clear in the ordering guidelines: “Images generated by this service are for personal entertainment and social sharing only, please do not use them for commercial purposes.” This protects me and reminds users.
At this point, you might think it’s just a grind.
But there’s a way to scale. When I started getting over 10 custom orders a day, I realized I couldn’t keep doing everything myself. I tried using Coze to set up a simple workflow where users could upload their photos, and AI would generate a basic version that I could refine. My efficiency tripled.
A bolder avenue is creating paid knowledge content.
After streamlining the process, I compiled my “hot keyword library” and “pitfall guide” into a 39 yuan resource pack. To my surprise, more people bought that than sought custom orders. Many prefer the satisfaction of doing things themselves over relying on others.
The best products often emerge from solving your own problems and crafting that “shovel” along the way.
Looking back at this project, it’s no longer just about “generating an image.” It’s a complete practice run of a micro-business loop: from demand insights, product packaging, traffic acquisition, sales conversion, to delivery optimization and scaling.
You might ask if I’m still doing this. Yes, but my focus has shifted. Now, I’m more like an “arms dealer,” providing “weapons” and “maps” to newcomers looking to enter the field. The potential here is far greater than just editing images one by one.
Transforming the “sweet” of the virtual world into real growth in our actual accounts—let’s do it together!
Top comments (0)