BEYOND
THE EDGE.
AI understands the complexity of the production field.
Transforms industrial data into actionable intelligence.
How It Works
The AutoBrics Platform Ecosystem
Connecting the factory floor, your data, and institutional knowledge into one intelligent flow - from the camera on the line to the decision on your desk.
INPUT
AutoVision H/W
Modular edge-AI hardware that adapts to your floor conditions and runs high-speed vision inspection at line speed.
INFERENCE
Inspection Workspace
Zero-shot domain models deliver high-precision defect detection and real-time visualization - no additional training required.
PIPELINE
Data Refinery
A GCD-based pipeline automatically sorts and cleans images and metadata into a production-ready dataset.
LEARNING
Knowledge Base
Process history and inspection decisions are consolidated, so performance improves with every run.
A closed loop: the Knowledge Base feeds back to push inspection accuracy higher with every roll.
Live Inspection
See every defect the moment it appears.
Zero-shot domain models catch knitting, weaving, and dyeing defects at line speed - needle lines, stains, and holes are mapped onto a live web view with confidence scores.
- No per-line training - a new fabric is running within minutes.
- Human-in-the-loop review is built into the flow, not bolted on.
Smart Data Engine
Every image labeled. Every image an asset.
The GCD refinery pipeline sorts and cleans images and metadata automatically - and surfaces defect types nobody has seen before, so the dataset grows on its own.
- GCD classification accuracy
- 0%
- Quality history per roll
- QR
Trusted Quality
Objective quality, on standards buyers already trust.
Grading follows the 4-Point System and ASTM D5430 - the number on the certificate means the same thing on your floor and to a buyer on the other side of the world.
- Automatic scoring - no manual re-inspection.
- Certificates you can share and trace, roll by roll.
The Hardware
AutoVision R2R System
A high-performance line-scan vision system built for roll-to-roll web inspection. Light-blocking enclosure, motorized camera bar, and precision lighting in a single industrial unit.
Fixed at 523 mm working distance to cover the full fabric width in one pass.
The enclosure blocks ambient light, holding image quality steady on the floor.
PWR · RUN · ALM states are readable at a glance from the line.
Precision inspection sustained at 50 m/min line speed.
OPTICS
4×
Line-scan cameras · 523 mm WD
THROUGHPUT
0 m/min
Inspection at line speed
FOOTPRINT
0mm
Installed width (incl. base)
DEPLOYMENT
Retrofit
Mounts onto existing lines
Our Roadmap
From technology to proven value.
From AI inspection automation to data as an asset, and on to global commercialization - one stage at a time.
2025
Foundation & AI inspection automation
AutoVision hardware, zero-shot anomaly detection, and a real-time inspection UI with human-in-the-loop review.
KPI · AUROC 96% / 50 m/min
2026
Deeper intelligence & data as an asset
GCD-driven discovery of new defect types, a QR-based quality history database, and an SOP-grounded knowledge agent.
KPI · GCD accuracy 90%+
2027
Commercial scale & global reach
Fabric optimization agents, a Southeast Asia base in Vietnam, and a 10% cut in fabric waste through process optimization.
KPI · KRW 550M cumulative revenue
Our Journey
How far we have come.
One year since founding - a record of AutoLabs' growth through technology, partners, and investment.
2025
Foundation & technical validation
MAY
AutoLabs Inc. founded
The official start of our industrial vision AI business.
JUN
Patent filed for AI-based automatic fabric defect inspection
Securing intellectual property on our core inspection algorithms.
NOV
MOU with Inha University SW-Centered University Program
Laying the groundwork for research and talent partnerships.
NOV
In-house foundation model published at BMVC 2025
Our vision foundation model research accepted at the international BMVC 2025 conference.
2026
Data as an asset & global expansion
JAN
MOU with Daesung TS, Ho Chi Minh City, Vietnam
A partnership secured for on-site validation in Vietnam.
MAR
Named top company in the Pre-Startup Package
Recognized as the highest-performing company in the program.
MAR
Joined the Antler Korea program
Selected for the accelerator cohort of global early-stage investor Antler.
APR
Patent filed for automatic defect clustering and classification
Securing IP on the core technology that clusters and classifies defect types automatically.
APR
Selected for Google for Startups
Joining Google's startup support program.
JUN
Pre-seed investment from Antler Korea
Raised pre-seed funding from global early-stage investor Antler Korea.
JUN
Selected as an NVIDIA Inception Partner
Recognized as an NVIDIA Inception partner for edge AI development.
JUN
Selected for the Majung Program (Microsoft)
Establishing a foundation for collaboration with Microsoft for Startups and other global partners.
Meet the Team
Making industrial AI real.
AutoLabs starts from real industrial problems and builds AI that is proven where it matters - on the floor.
Junha Yoo
Chief Executive Officer
David Kim
Chief Operating Officer
Yunseok Oh
Chief AI Officer
Hwichul Shin
Chief Product Officer
Backed by
The partners beside us.
Global investors and technology partners are with AutoLabs on this journey.
Request a PoC
See AutoBrics running on your own line.
Request a PoC and watch real-time defect detection and objective grading work on your own fabric.