How Chance AI Is Taking On Google Lens With a Curiosity-First Vision Agent, Explains Dr. Zeng

The artificial intelligence (AI) revolution has altered the trajectory of technology, and its impact ripples through industries, companies, institutions, and individuals. Capabilities that would have been as fantastical as a flying car or a working lightsaber, even just 10 years ago, are now available to the masses for free. And behind these tools and products are major tech giants and startups founded in the last three years, with names ending in “AI”.

This major shift in the tech landscape has also had a democratising effect, with companies with market valuations matching a country's GDP now directly competing with startups with 11 employees, working out of makeshift office space in an apartment. While some experts believe it is just a temporary bubble driven by investor excitement, others feel this David vs. Goliath battle is here to stay.

Among the latter is Dr Xi Zeng, a product design veteran whose long career includes directorial roles at OnePlus and its parent Oppo, as well as at ByteDance (the company behind TikTok), and a five-year stint as a professor at the China Academy of Art. After overseeing products purchased and used by millions for over a decade, the AI revolution also pushed Zeng to chase the entrepreneurial dream, and in June, he founded Chance AI.

So, how does one go from working on smartphones and social media apps to leading a tech startup and developing a product that aims to directly compete with Google Lens? Gadgets 360 recently interviewed Dr Zeng to find the answers to these questions and delve deeper into the startup's AI app.

How Does an Employee Become an Entrepreneur?

When Zeng reflects on why he walked away from senior roles at ByteDance and OnePlus to build something of his own, he does not describe a single lightning-bolt moment. Instead, he mentioned that it was a desire that grew stronger over time. “It's been more than a decade working with ByteDance,” he says. “I have seen a product being built from scratch to serve more than 100 million users. I found it very exciting to build products. This is where my passion is.”

That passion, he admits, had begun to feel boxed in. In a giant organisation, he says, “They care more about the infrastructure than the consumer focus.” The rise of generative AI only sharpened this realisation. After models such as GPT arrived, he saw an industry rewiring itself overnight. “Maybe in the past you needed 100 or 200 developers to build something you could ship,” he says. “Right now, it is only five or six developers with the help of AI.”

With that, realisation began Zeng's entrepreneurial journey. However, just like Rome was not built in a day, neither can a product that intends to compete against multinational corporations. But there was one moment of frustration that shaped the product's direction.

The Inception of Chance AI: Curiosity Lens

Years ago, while backpacking across Europe, he stood before the Sagrada Familia, the architectural masterpiece designed by Antoni Gaudí in Barcelona, Spain. Wanting to learn more about the structure and its history, he pulled out his phone and opened Google Lens. The encounter stayed with him.

“I was frustrated,” he said, adding, "I found that the result that I got wasn't very satisfactory. It was all shopping links guiding you to buy a souvenir or to some e-commerce website. My curiosity wasn't satisfied with the shopping list. I wanted to know more about the thing in front of me, not where to buy something.”

Chance AI Founder Dr Xi Zeng
Photo Credit: Chance AI

That disconnect, between the human desire to understand the world and the digital economy's desire to sell it, became the genesis of Chance AI, explained Dr Zeng. In a market dominated by text-based large language models (LLMs) and transaction-focused image search, Chance AI is betting on a different philosophy: that the future of AI isn't about productivity, but curiosity.

Chance AI positions itself as a visual-first agent. “Today's large language models mostly focus on text,” he explains. “It's a game of predicting the next word.” That makes them powerful for productivity tasks, but narrow in context. “Vision,” he says, “is the most intuitive way for people to interact with the world. It's the human operating system.”

Once the AI analyses something, taking an action on it is the agent's job. A pair of sneakers becomes a price comparison. A concert flyer becomes a weekend plan. “There's always an action coming after what you see. That's what we call an agent.”

Battling the AI Giants

The decision to build Chance AI required leaving the safety of established tech hierarchies. "Maybe in the past you needed like more than 100 or 200 developers to build something you can ship, but right now it is only like five or six developers with the help of AI," Zeng notes.

More importantly, he felt that the "giant organisations" were too focused on infrastructure and commerce to build a tool purely for visual intelligence. "They want you to pay your money," he said of competitors like Google Lens. "But this is something we don't encourage. We want you to learn. We want you to be inspired."

Privacy, Boundaries and the Realities of Scaling

The technology behind Chance AI relies on open-source vision language models (VLMs) ranging from three billion to 74 billion parameters, which the team post-trains to suit specific user needs. However, building a camera-first platform in an era of data privacy laws such as GDPR comes with significant responsibilities.

Every visual-driven product must wrestle with sensitive data. Zeng is careful to clarify how Chance works behind the curtain. “First of all, the data is stored in the local market,” he says. “Images are extremely sensitive, and everything is encrypted.” They rely on AWS in India, Southeast Asia and Europe, and Google Cloud in the Americas. He also maintains that images uploaded by users are not seen or used by the company for any reason.

However, images posted publicly within the Chance community are visible by design. “That's something users decide to publish.” When it comes to harmful or inappropriate uploads, Zeng says the app draws a firm line. “The first three seconds after you upload a photo, it's calculating whether it is appropriate to be delivered.” Repeated violations trigger warnings. “After five attempts, we sent a notice. If they keep doing that, we block. But we haven't had to do that to anybody yet.”

Monetisation, India Plans, and Intent

Chance AI plans to offer a premium tier, business-to-business (B2B) licensing and possibly a recommendation-driven marketplace. But he is quick to differentiate Chance from traditional behaviour-shaping algorithms. “Recommendation algorithms are a dying technology,” he says. “They are still controlled by marketing developers behind the scenes. Today, the model actually understands the image you upload. It understands you better than the algorithm.”

The company recently reached 2,00,000 monthly active users (MAUs), with significant traction from India, a market the team didn't originally plan to enter. “It was karma,” he smiles. “We already had a bunch of young, curious minds in India using our product every day. The expansion was just following nature.”

His years at OnePlus also influenced his decision. “OnePlus has a strong community DNA and a strong bond with the Indian market,” he says. “It's not my first time seeing the creativity of Indian users.”

The Five-Year Horizon

Looking ahead, Zeng sees Chance AI evolving beyond a smartphone app. He predicts a shift in hardware that will make visual AI the dominant interface. He reasons that people using a tiny screen and typing as an interface are not getting an immersive enough experience in the modern world. He envisions a future of AI glasses where the visual agent acts as a companion.

Zeng believes this shift is inevitable because of the generational divide. While older generations rely on text and specific search queries, "Gen Z" is visually native, growing up on Instagram and TikTok. "I am 100 percent sure it's not going to be a [bubble]," he asserts regarding the visual AI boom.

For now, Chance AI remains focused on its primary mission: refining the "search engine" for the visual world. But the ultimate goal is to transform how we process reality, turning every glance into an opportunity to know, rather than just a chance to buy. As Zeng puts it, "Curiosity is the purest rebellion you can do in your world."