Collov Labs, an AI lab advancing foundational visual AI technology, today launched NewEyes AI, an app built on the company’s proprietary visual AI platform, delivering the most advanced visual AI on the market today in an app that allows users to point their smartphone camera at a scene and have AI agents take the next step for them, turning a photo into a finished task rather than just an answer.

The launch follows Collov Labs’ $23M Series A earlier this year, which the company used to advance its visual AI research and platform capabilities. NewEyes AI is the first product built on that platform since the raise, reflecting the lab’s thesis that images – not LLMs – will bring AI to the masses, and Collov Labs’ platform will enable anyone to take advanced actions from their phone and eventually, other wearables like glasses.

“For most people, the camera is the most natural way to show technology what they’re dealing with,” said Xiao Zhang, CEO and co-founder of Collov Labs. “Once that camera can reliably understand a scene – whether it’s a room, a document, or a routine – the same foundation can support experiences on today’s smartphones as well as tomorrow’s wearables.”

Most of the visual AI technology that exists today is built around visual question answering (VQA), a system that can answer questions about objects in an image, but stops there. Collov Labs’ proprietary agentic visual AI system goes beyond this. The agents convert photos into structured, queryable representations and identify objects, materials, spatial layout, and then execute next steps based on the app’s understanding of the user’s personal world model – what’s already in their closet, pantry, or plant collection. This allows NewEyes AI to reason about each new photo in light of everything it has seen before, as well as the user’s stated goals, enabling multi-step actions such as:

  • Snap your closet, skip the duplicate purchase. NewEyes AI remembers what you own and only recommends what’s missing.

  • Snap a pitch deck, walk away with a website. Fully reconstructed and hosted before you leave the room.

  • Snap a parking sign, avoid a ticket. NewEyes AI reads the fine print and reminds you before time’s up.

Other visual AI technology, such as Google Lens, are fundamentally recognition tools, returning a match for something else to act on. NewEyes AI works differently: recognition is the trigger itself, autonomously selecting and running the right skill in the same reasoning pass — straight through to action, with no hand-off between a ‘vision’ step and an ‘agent’ step.

NewEyes AI is built on the same foundational technology that powers Collov Labs’ two other apps, Collov AI and CozyAI, along with the usage data those apps have generated from over two years being live with now more than 1 million users, 10,000 paying customers, and 300 enterprise collaborators. That base of real-world visual interaction – much of it from non-technical users who quickly became power users – has been fed back into the platform’s retraining loop to refine how NewEyes AI recognizes scenes and follows through on multi-step requests. Other use cases include:

  • Breakdown the nutrition and macro breakdown of a plate

  • Identify plant species with recommended care routines

  • Get pet information with a breed and health analysis

  • See a pros-and-cons analysis with sourced comparisons of products

  • Learn more about the history of art and the techniques of artists

  • Translate text, extract text, or identify a landmark from any photo

  • Redesign a room in a specific style and compare furniture prices

  • Compare prices of products across retailers with links to buy

  • Get explanations, bug fixes, and suggested improvements to code

“When we built Collov AI and CozyAI, we saw people who’d never even touched ChatGPT become power users within days,” said Casey Zhou, CTO and co-founder of Collov Labs. “NewEyes AI takes that same system and makes it general-purpose.”

The system is built to keep improving on its own. Every interaction is converted into a structured training signal that feeds back into the model, the agent’s reasoning policy, and its internal validators, so the platform gets more accurate the more it’s used. That closed-loop retraining runs on top of a proprietary, in-house-built diffusion transformer tuned specifically for image and scene generation and editing.

NewEyes AI is available first as a mobile app, and in the months ahead, Collov Labs plans to extend the same camera-first experience to wearable devices, and the company’s underlying visual intelligence system is currently in beta testing with several existing AI glasses and wearables, including Meta’s Ray-Ban line.

About Collov Labs

Collov Labs is an AI research and product company building the intelligence layer for visual AI. Founded in 2021 and headquartered in Silicon Valley, the company develops visual intelligence systems that combine diffusion models, spatial reasoning, and agentic workflows — enabling AI to understand scenes, reason about visual context, and translate images or camera input into real-world actions. Its visual intelligence model was one of the first consumer applications directly integrated into Qualcomm’s on-device system and is featured in the Qualcomm® Hexagon™ NPU. Collov Labs’ products, Collov AI, CozyAI, and now NewEyes AI, have grown to over 1 million users, 10,000+ paid customers, and 300+ enterprise collaborators across retail, design, and e-commerce.

Media gallery

About The Author
Business Wire