Skip to main content

Tech Veteran Launches liko.ai to Bring Smarter Privacy-Focused Home Cameras

A New Vision for Home Security

Tech industry veteran Ryan Li is shaking up the smart home market with his new startup liko.ai, backed by major players including SenseTime and iFLYTEK. The company promises to deliver home cameras that actually understand what they're seeing - without compromising your privacy.

Solving Camera Pain Points

Today's home security cameras suffer from two major flaws that liko.ai aims to fix:

  • Privacy risks: Most devices upload footage to the cloud, creating potential vulnerabilities
  • Limited intelligence: Basic motion detection hasn't evolved much beyond its initial implementation

The solution? Keep everything local. "We process all video on the device itself," explains Li's vision. "Your living room footage should stay in your living room."

Smarter Than Your Average Camera

The company's secret sauce lies in its lightweight visual language model that runs directly on devices. Unlike traditional cameras that simply detect movement, liko.ai's system can interpret scenes semantically:

  • Recognizing when a child approaches stairs unsupervised
  • Identifying pets causing mischief
  • Answering natural language queries like "Who entered the study yesterday afternoon?"

Building the Home AI Hub

liko.ai envisions its products becoming central hubs for home intelligence:

Edge Computing Power: Combines visual understanding with local processing chips Expandable Ecosystem: Future versions will connect with smart locks, lights and other devices Discreet Design: Products avoid the "security camera" look to blend into modern homes

The technical approach draws on Li's hardware expertise from building Meituan's ecosystem and leading Lexar Global.

Industry Backing Speeds Development

With support from SenseTime (computer vision specialists) and iFLYTEK (speech AI leaders), liko.ai expects rapid progress. Insiders hint at a mid-2026 launch targeting premium home security and childcare markets.

The $10 billion global home camera market hasn't seen fundamental innovation in years. As one industry observer noted: "We've had cameras that see - now we're getting cameras that understand."

Key Points:

  • Former Meituan exec launches privacy-focused AI camera company
  • Edge-based processing eliminates cloud privacy concerns
  • Advanced visual models enable semantic understanding of scenes
  • Backed by major AI players SenseTime and iFLYTEK
  • First products expected mid-2026

Enjoyed this article?

Subscribe to our newsletter for the latest AI news, product reviews, and project recommendations delivered to your inbox weekly.

Weekly digestFree foreverUnsubscribe anytime

Related Articles

News

NVIDIA's Lyra 2.0 Turns Single Photos into Expansive 3D Worlds

NVIDIA has unveiled Lyra 2.0, a groundbreaking AI system that transforms ordinary photos into detailed 3D environments spanning up to 90 meters. The technology addresses a major challenge in virtual world creation - maintaining consistency across large spaces. What sets Lyra apart? It remembers spatial details like a human would, correcting errors on the fly while outperforming competitors in visual quality. The system already integrates with NVIDIA's Isaac Sim, offering new possibilities for robot training and autonomous vehicle development.

April 17, 2026
NVIDIA3D generationAI innovation
Ant Group's Lingbo Tech Open Sources Breakthrough 3D Mapping Tool
News

Ant Group's Lingbo Tech Open Sources Breakthrough 3D Mapping Tool

Ant Group's Lingbo Technology has made waves by open-sourcing its revolutionary LingBot-Map, a system that creates real-time 3D reconstructions using just a standard camera. Unlike previous methods that required specialized equipment or post-processing, this innovation works on the fly during video capture, achieving impressive 20FPS performance. The technology promises to transform fields from robotics to AR by making high-quality spatial mapping more accessible than ever.

April 16, 2026
3D reconstructioncomputer visionAnt Group
Tencent's Breakthrough Video Tech Speeds Up Generation by 11.8 Times
News

Tencent's Breakthrough Video Tech Speeds Up Generation by 11.8 Times

Tencent's Hunyuan team has cracked the code on slow video generation with their new DisCa technology, achieving an impressive 11.8x speed boost without sacrificing quality. This open-source solution, accepted by top computer vision conference CVPR 2026, introduces smart feature prediction that revolutionizes how AI creates videos. The team also improved upon MIT's approach to make it work better for complex video tasks, with results already powering their latest video generation model.

April 16, 2026
AI video generationTencent researchcomputer vision
JD.com Unveils Cutting-Edge AI Training Camera for Next-Gen Robotics
News

JD.com Unveils Cutting-Edge AI Training Camera for Next-Gen Robotics

JD.com has introduced the JoyEgoCam, a groundbreaking data collection device designed to train AI systems through real-world observation. This industrial-grade camera captures ultra-high-definition footage at 60 frames per second, enabling machines to learn subtle movements and environmental changes. The launch comes as part of JD's ambitious plan to collect 10 million hours of video data within two years, potentially transforming warehouse automation and logistics robotics.

April 16, 2026
AI trainingroboticscomputer vision
Google's AI Breakthrough Teaches Machines to See Like Humans
News

Google's AI Breakthrough Teaches Machines to See Like Humans

Google DeepMind has cracked a major challenge in AI vision with its new TIPSv2 system. While current models can describe images broadly, they stumble on fine details - like locating a panda's left hind leg. The solution came from an unexpected finding: smaller models sometimes outperform larger ones in segmentation tasks. By refining training methods and reducing computational overhead, TIPSv2 achieves 14% better segmentation accuracy while using 42% fewer parameters. This breakthrough could revolutionize fields from medical imaging to autonomous vehicles.

April 16, 2026
computer visionmachine learningAI research
Microsoft's New AI Model Packs a Punch with Smart, Lightweight Design
News

Microsoft's New AI Model Packs a Punch with Smart, Lightweight Design

Microsoft has unveiled Phi-4-reasoning-vision-15B, a surprisingly powerful yet lightweight AI model that excels at visual reasoning tasks. What makes it special? It delivers top-notch performance while keeping computing costs low, making it ideal for resource-constrained environments. The secret sauce? High-quality training data and an innovative hybrid reasoning approach that automatically adjusts to simple or complex tasks. Now available as open-source, this model could change how we think about efficient AI.

April 13, 2026
Microsoft AImultimodal reasoningefficient AI