Bronze Medal Winner - Geneva Inventions 2024

CarryAI - Advanced Vision-Language Model AI Solutions

Perceive · Reason · Execute

Advanced Vision-Language Model Solutions for Enterprise Impact. Transforming visual data into measurable business outcomes with proven VLM precision.

Vision-Language Model AI for Facility Management, PropTech & Construction Tech

Our VLM technology fuses computer vision and natural language processing to deliver real-time site intelligence. Automatically detect defects, track progress, verify compliance, and generate audit-ready reports from any image or video feed—reducing downtime, mitigating risk, and accelerating project handovers.

Why CarryAI's Enterprise VLM Stands Apart

Engineered for mission-critical operations, our VLM delivers unmatched privacy, speed, and reliability—fully on-device, zero cloud dependency, and purpose-built for regulated industries.

• Edge Deployment (100% On-Device) — Runs entirely on Apple Silicon and edge devices. No cloud required. Your data stays on your infrastructure.
• Lightning Fast Processing (<3 Seconds) — From image capture to actionable result in 3 seconds or less.
• Complete Privacy (Zero Cloud Upload) — No images or videos transmitted to the cloud. Full data sovereignty and compliance.
• Optimized Intelligence — Brain capacity equivalent to a 4-7 year old child. Efficient, focused, and purpose-built for real-world tasks.
• Stable Output Rate (99%+ Consistency) — Consistent, reliable performance across different environments and conditions.
• Dynamic Prompting (Context-Aware) — Intelligent prompt adaptation for different scenarios and locations.

VLM AI in Action

See how our Vision-Language Models transform industries with practical, deployable solutions that deliver measurable results.

• Safety Monitoring — Real-time PPE detection, pose recognition, and hazard identification for construction sites and industrial facilities.
• Quality Control — Automated visual inspection for manufacturing with defect detection and classification at scale.
• Healthcare Analysis — Medical image interpretation, pathology analysis, and patient monitoring through visual AI.
• Retail Intelligence — Visual product search, inventory management, and customer behavior analysis for e-commerce.
• Autonomous Systems — Scene understanding and navigation for autonomous vehicles and robotics applications.
• Security & Surveillance — Intelligent monitoring with anomaly detection, crowd analysis, and incident reporting.

AI Safety Solutions

Our AI Video Analytics Solution implements advanced Object Detection and Human Pose Recognition to monitor worker status in construction sites and industrial facilities. Winner of Bronze Medal at Inventions Geneva 2024.

• AI Deep Learning with high accuracy detection
• 18-point skeleton reconstruction for pose analysis
• Multiple simultaneous pose detection
• Real-time monitoring and instant alerts
• Edge computing for fast, reliable processing

Our Vision

We are a small team of applied AI experts and researchers, half of whom are graduates of HKUST. We have core technology in AI Object Detection, Human Pose Recognition, and Large Language Models, and are committed to helping businesses and organizations harness the power of AI in their existing projects and solutions.

Our team is passionate about simplifying the AI adoption process, making it accessible and user-friendly for everyone.

Why Choose CarryAI?

We stand out from competitors with our unique approach to edge computing, customization, and unbeatable value proposition.

• Edge Computing Focus — Visual analytics at the edge using NVIDIA Jetson devices, without cloud dependency.
• Customizable Solutions — Tailored solutions designed for your specific needs.
• Computer Vision Expertise — Deep expertise in computer vision and AI.
• Complete Service — From model training and fine-tuning to integration and deployment.
• Unbeatable Pricing — Standard tier with clear ROI expectations.
• No Cloud Reliance — All processing runs completely on edge devices.
• Portable & Low Power — Lithium-ion battery packs for continuous 10-hour operation.
• Strong Customer Support — Ongoing assistance and training resources.

In-Depth Safety & Facility Management

Autonomous robotic patrols powered by VLM AI for continuous facility monitoring, safety compliance, and proactive maintenance.

AI-Powered Safety Monitoring

• Scheduled Patrols — Strict schedule-based inspections of emergency power boxes, performed hourly and on-demand
• Waste Check — Verifies trash and recycling bins have properly closed lids
• Security Check — Confirms doors are fully closed and secured
• Damage Patrol — Detects broken glass, cracks, dents, burnt-out or flickering lights
• Spotting Danger — Scans for leaks, puddles, and unusual spills on floors
• Critical Equipment Status — Verifies fire extinguishers, AEDs, and first-aid kits are accessible

Advanced Robot Capabilities

• Event Detection Platform — Centralized platform for searching and reviewing all detected events
• PTZ Camera System — Pan-Tilt-Zoom camera for investigating detected events
• Checkpoint Patrol — Autonomous patrol along predefined routes
• Weather Resistant — Rain and splash-proof design
• Auto-Return Charging — Automatic return to charging dock for 24/7 operation
• Obstacle Avoidance — Advanced sensors and AI-powered navigation

VLM + CCTV & Robotic Patrol Integration

By seamlessly integrating CarryAI VLM (Vision-Language Model) with existing CCTV solutions and robotic patrols, traditional surveillance is transformed into a proactive, intelligent security network. This powerful combination enables 24/7 precise detection and analysis of various safety and environmental anomalies across public and managed areas:

• Behavior & Order Management: The system automatically flags unauthorized activities, such as cycling or scooter riding in restricted areas. It also monitors individual well-being by detecting anomalies like sitting, sleeping, or when a person has fallen and may require assistance.
• Security & Threat Detection: With an advanced understanding of human behavior and objects, the VLM provides split-second alerts for aggressive behavior—including fighting and situations where people are being hurt or harmed. Concurrently, it can identify dangerous items such as knives, firearms, and other attacking equipment.
• Environmental & Sanitation Hazards: The solution acts as an immediate line of defense by identifying fire, smoke, and flammable dangers, as well as localized flooding or heavy water accumulation. It even extends to micro-level estate management by spotting uncleaned dog feces.

By blending the static overhead view of CCTV with the dynamic, mobile presence of robotic patrols, CarryAI VLM delivers human-like visual comprehension to security systems, drastically cutting down response times and elevating property management standards.

VLM AI + Robotics Integration

Combining Vision-Language Models with cutting-edge robotics platforms for autonomous intelligence and real-world applications.

Exhibitions & Events

CarryAI has showcased cutting-edge VLM AI and robotics solutions at major exhibitions and industry events worldwide, demonstrating real-world applications and innovations.

Ready to Transform Your Business?

Let's discuss how CarryAI's Vision-Language Model solutions can help you achieve your goals. Our team is ready to understand your unique challenges and develop custom AI solutions.

Schedule a Consultation | WhatsApp Us