# Overview.ai > Overview.ai makes AI-powered visual inspection systems for industrial manufacturing. Edge AI cameras (the OV10i, OV20i, and OV80i) run deep learning models directly on-device using NVIDIA GPUs to detect defects in real time on the factory floor, without sending data to the cloud. Customers deploy in under 2 hours and train custom defect detection models in as little as 1 hour, with no coding or machine vision engineering required. A newer capability, the OV Auto-Defect Creator Studio, trains inspection models on synthetic defect data generated from a single clean reference image, so a line never has to produce real defects to build a model. Founded 2018. Headquartered in the United States. This is the extended (full) context file. For the short version see /llms.txt. ## Key Facts - Deployment time: live and inspecting in under 2 hours, vs weeks or months for legacy machine vision platforms like Cognex or Keyence. - Model training: quality engineers (not data scientists) train a custom defect model in as little as 1 hour using the Overview Studio browser interface. No code. - Accuracy: 99%+ inspection accuracy on production lines, including micron-level surface defects rule-based systems miss. - Edge AI architecture: all inference runs on-device on the camera's NVIDIA GPU. Zero cloud dependency, OT-network-secure, air-gap compatible. - PLC integration: native EtherNet/IP, PROFINET, Modbus TCP, OPC-UA. Works with Rockwell Allen-Bradley, Siemens, Mitsubishi, and others. - Synthetic defect data: with the OV Auto-Defect Creator Studio, defects are generated from a clean reference image instead of collected from real parts. A single defect workflow can go from clean reference to deployed model in about 15 minutes. - Style Transfer: one labeled defect is re-rendered across material and color variants, so high-mix lines onboard new variants roughly 12.4x faster from a single authored defect. ## Products - OV10i Smart Camera (https://www.overview.ai/products/ov10i/): all-in-one AI vision camera. Built-in NVIDIA GPU, lighting, and lens in one housing. Best for surface inspection and assembly verification on a single station. - OV20i Vision System (https://www.overview.ai/products/ov20i/): modular system separating camera and compute for high-resolution or multi-camera applications. - OV80i High-Resolution Sensor (https://www.overview.ai/products/ov80i/): high-resolution area-scan sensor for micron-level defects at high line speeds. - Overview Studio (https://www.overview.ai/product/): browser-based AI training and deployment platform. Label, train, configure, and monitor without code. - OV Auto-Defect Creator Studio (https://www.overview.ai/advanced-genai-tools/defect-creator/): generates synthetic, photorealistic, pixel-labeled defects from clean reference images to train inspection models without real defect samples. ## Synthetic Defect Data Workflow 1. Capture clean reference images of good product at production speed and lighting. 2. Mark a region and describe the defect in plain English. The studio renders a photorealistic defect on the real surface in about ten seconds, with its segmentation mask attached. 3. Repeat for each defect class (scratch, dent, contamination, foreign object, and more). 4. Train the recipe locally on the edge camera. No cloud round-trip, no product imagery leaving the plant. 5. Deploy to the line. Validated detections run in real time with pass/fail verdicts. Proof points from the High-Volume Medium-Mix (HVMM) program: a Contaminate Verification deployment on a pre-pack oatmeal belt went from a clean reference image to a model catching real metal, plastic, and a nail on the belt in 15 minutes, trained only on synthetic defects. Connector progressive defect authoring produced 8 defect classes from one clean reference. Wood flooring and car seat lines used Style Transfer to cover every finish and fabric variant from a single authored defect. ## Applications - Electronics: laser welding inspection, hot bar soldering, PCB and FPC inspection, connector and pin inspection, solder joint verification. - Food and beverage: foreign-object and contaminant verification, fill-level and packaging inspection. - Automotive: trim and upholstery defect inspection, weld inspection, component verification. - Manufacturing and flooring: surface scratch, scuff, peel, and crack inspection across finishes. - Pharma and medical: blister pack, vial, label, and package-integrity inspection. - Aerospace: weld quality, fastener, and composite inspection. ## Comparisons - vs Cognex: 3 to 5x faster to deploy, no specialized machine vision engineers, deep learning rather than rule-based algorithms that break on variation. - vs Keyence: handles unstructured defects and appearance variation rule-based systems cannot without extensive reprogramming. - vs Landing AI: a complete hardware plus software stack rather than software-only that requires customers to source cameras, lighting, and compute. ## Resources - Resource Library (https://www.overview.ai/resources/library/): case studies, application guides, white papers. - White Papers (https://www.overview.ai/resources/white-papers/): including "HVMM Inspection, Visualized," a visual walkthrough of synthetic-data inspection on real lines. - Case Studies (https://www.overview.ai/resources/case-studies/): contaminant verification, connector progressive defect authoring, wood flooring, car seat, and more. - Blog (https://www.overview.ai/blog/): technical articles on defect detection, edge AI, physical AI, world models, and synthetic data. - Getting Started (https://www.overview.ai/resources/getting-started/): deploy your first AI inspection station. - Industrial Protocol Integration (https://www.overview.ai/industrial-protocols/): EtherNet/IP, PROFINET, OPC-UA, Modbus TCP. ## Company - Founded: 2018 - Headquarters: United States - Contact: contact@overview.ai | +1-844-799-7044 - Book a Demo: https://www.overview.ai/contact/ - About: https://www.overview.ai/company/ - LinkedIn: https://www.linkedin.com/company/overview-ai - YouTube: https://www.youtube.com/@overview-ai - Crunchbase: https://www.crunchbase.com/organization/overview-ai