Chance AI raises millions for visual AI agent topping 86% accuracy
What's the deal? Chance AI, a visual agent startup founded in 2025, has closed an angel round worth several million dollars. Meitu led the round, with NYX Ventures and Alibaba-affiliated investors joining. The funds will go toward model improvements, North American growth, and early commercialisation.
The company's core product is a camera-first AI tool — instead of typing prompts, users point their phone camera at something and the system interprets visual intent, offering judgements, suggestions, and action plans. Founder Zeng Xi previously worked at OnePlus, OPPO, and ByteDance, where he helped build AI products including Doubao.
Why now? Chance AI's visual agent hit 86.07% accuracy on the MMMU-Pro multimodal reasoning benchmark — ranking first globally and surpassing the human baseline of 85.4%. The product has reached roughly 200,000 users across 35-plus countries, with about 40% in North America. It twice topped Product Hunt's daily chart.
The startup targets young women on North American university campuses — a demographic it calls "visual natives," raised on Instagram, TikTok, and selfie culture. Core use cases include personal style analysis, outfit checks, design feedback, and social expression. Chance AI has already built networks at NYU, USC, and other schools, using campus events to acquire seed users.
What could go wrong? Retention will be the real test. The company reports a 30-day revisit rate of 49.2%, which it says leads the industry — but sustaining that as it scales beyond early adopters is another matter. Its revenue model remains nascent: premium subscriptions, hardware licensing, and cautious ad recommendations are planned, but the current priority is habit formation, not monetisation.
Camera-based AI also raises privacy concerns. A product that continuously interprets what users see — and builds a personal "visual memory" of style preferences, wardrobe, and social image — will need to earn deep trust, especially among younger users.
The signal: Chance AI reflects a broader shift in how AI products think about input. The text box has dominated the first wave of consumer AI, but a growing cohort of startups bets that cameras, not keyboards, are the more natural interface. Meitu's backing reinforces this — the Chinese imaging company sees visual AI evolving from simple recognition toward understanding taste and aesthetics.
The startup's campus-first distribution strategy also echoes early playbooks from social platforms like Facebook. If Chance AI can convert its visual agent from a utility into a lifestyle community — where users share AI-generated content and interact around style — it could carve out a durable niche in a crowded consumer AI market.
Read more: 36kr.com
