Reported by NUPIAO |
Edited by NUPIAO | Shuqi Wen
There’s a brand-new AI app taking over North American campuses, and the craziest part? It doesn’t even have a single input box.
Open up Chance AI and you’re greeted with just one button: a camera. Point it at that questionable top you’re debating wearing out, and within seconds, the app breaks down the vibe, even cross-referencing that blue hat you rocked last week with a quick “pair it with this instead.” Stare at an unfamiliar painting? It’ll instantly unpack Picasso’s Guernica, the bombing that inspired it, and why it still hits different today.
Founded back in 2025 and headquartered in Singapore with a core R&D team based in Shenzhen, the startup just announced a multi-million dollar funding round last May. Meitu led the round, with follow-ons from NYX Ventures and investment arms tied to Alibaba. By then, the app was already cruising near the top three on the App Store’s free charts, boasting over 200K users across 35+ countries and a staggering 49.2% monthly return rate.
The brains behind the operation, Dr. Zeng Xi, cut his teeth studying cognitive science and contemporary art at the University of Barcelona. He spent years exploring how humans use visuals to make sense of the world. After grad school, he dove into product management at OnePlus, shaping the early hydrogen OS and hardware roadmaps from the 5 to the 9 series, before stepping up as Design Director at OPPO.
In 2023, Zeng joined ByteDance’s Flow division to help birth AI products like Doubao from scratch. His takeaway? Almost every major AI right now forces you into a chat window because text prompts are literally closest to LLMs, which dominate answering questions and generating copy. But that’s not how we actually live our lives.
Enter Chance AI, where the keyboard goes to die. Zeng argues that chat boxes are fine for work, sure, but everyday life isn’t a spreadsheet. You usually spot something, get curious, let it sink in, and then decide what to do next. That’s exactly why he flipped the script and made the camera the primary gateway.
We crave multimodal interactions, and visuals are undeniably the centerpiece. Yet most current tools barely scrape the surface, simply labeling objects instead of actually interpreting them. As Zeng puts it, while the rest of the industry is stuck racing on language models and task bots, Chance AI is building a whole new AI layer rooted in sight—helping you read a scene, form a judgment, and take action.
“Human understanding follows four steps: eyes see, signals fire, travel through nerves, and finally, the prefrontal cortex kicks in,” Zeng explains. “Most companies try to mash those into one step. We split them up, giving each stage its own dedicated model and engineering polish.”
Upload a photo to Chance AI, and it immediately maps the context, figures out what you’re after, and routes the job to the right agent. The more you use it, the smarter it gets—building a personal visual profile tracking your style quirks, wardrobe staples, and social aesthetic. It keeps feeding you tailored insights and emotional resonance along the way.
On the hardcore MMMU visual benchmark, Chance AI currently sits at number one globally. It clocked an 86.07% accuracy rate, edging out human raters (85.4%) and every other mainstream giant in the room.
The app originally blew up across North American campuses. By hosting real-world meetups, Chance AI quickly built buzz at NYU, USC, Stanford, and beyond, slowly waking up the college crowd hanging around these schools. Once it cracked the campus code, strategic holiday drops pushed engagement through the roof, and user growth went parabolic.
Take Christmas 2025, for example. Tons of students used Chance AI to hunt down gifts, snapping pics of items and chatting with the AI to see if they’d actually land well with their friends. When fall rolled around, the “dorm decor reveals your personality” feature had everyone spamming group chats with their setups.
College girls were the earliest adopters, dubbed the “Visual Native” generation. Raised on Instagram feeds, TikTok scrolling, and selfie culture, they naturally process and express themselves through imagery. That’s why features like style audits, image analysis, fit checks, design critiques, and social expression became instant hits within the app.
Rivals are practically non-existent right now. Look at the global AI landscape: roughly eighty percent of Silicon Valley startups chase B2B contracts. Of the remaining twenty, many are locked into infrastructure plays. The few consumer apps out there mostly stick to voice memos and note-taking because cashing in is dead simple. Lifestyle-focused consumer AI? Barely anyone’s touching it. Why? Because early AI gold rushes favor signing enterprise deals that guarantee runway, rather than grinding months to grow a C-end user base.
Zeng sees this blind spot as Chance AI’s golden ticket. “Models are just electricity. The hard part is building the actual appliance. I’m really good at wiring power into usable tools.”
Right now, Chance AI stays completely free, just throttling daily uses to keep servers happy. Zeng confirms commercialization kicks off post-funding. The playbook? Offer premium subscription tiers for extra memory storage and advanced features, while keeping the core experience totally free-to-play.
So what makes people actually open their wallets? Vertical visual memory accumulation and industry-leading low hallucination rates. Beyond that, Zeng believes one word anchors the entire product philosophy: Meaning.
“Hang a painting on a wall, and any animal just sees cloth and chemicals. For us humans, though, it’s a snapshot of civilization,” he points out. “A random brick on the street and a brick from the exact path Da Vinci walked look identical. But the meaning behind them worlds apart.”
He bills Chance AI as an “AI-native visual narrator”—think movie recap channels, sports commentators, or museum docents—all designed to help you actually comprehend what you’re looking at. Once you master visual parsing, you hold the pen to interpret reality. “If I tell you that Supreme tee holds zero resale value, someone who’s never heard of the brand scans it with our app and genuinely nods along.”
For a category-defining product like this, every monetization experiment feels uncharted, which is great news and terrifying all at once.
Chance AI’s strategic backer, Meitu, might just hand them a playbook. Meitu successfully pivoted from ad-driven revenue to subscription-heavy models, pushing past 100M monthly active users outside mainland China, with growth outpacing domestic numbers. Since late 2025, most of their newly paying subscribers are rolling in from Europe, the Americas, and East Asia—high ARPU playgrounds.
Zeng shares that Meitu founder Wu Xinhong tested the waters personally, loved the trajectory, and pledged joint exploration into overseas aesthetic use cases.
That said, the road ahead isn’t risk-free. Big tech competition looms large. GPT-4o and Gemini already pack heavy multimodal vision skills, while domestic super-apps like Doubao are rapidly patching in visual upgrades. Against universal utility tools, niche apps need ironclad reasons to convince people to download and keep a standalone icon on their home screen.
Fast-forward a few years, and Zeng’s north star is building the AgentOS for next-gen terminals. He views smartphone apps as just a transitional phase. Chance AI is engineering a visual brain that’ll seamlessly plug into AI glasses, smart wearables, and connected home devices once those form factors mature.
Hardware development? Not on the near-term radar. “The reason I’m skipping hardware might just be because I actually know how it works,” Zeng admits. “The supply chain isn’t stable yet. We’re waiting for Apple, Huawei, and the rest to drive costs down. Once the ecosystem matures, we’ll pick the optimal form factor—instead of forcing a shape onto an immature supply chain.”
Until that horizon arrives, visual AI is finally getting hit with its most critical test. And Chance AI looks ready to pass.
