Averroes

What it is

Averroes is a no-code AI visual inspection platform for manufacturing quality teams. It does one job, finding defects in images of parts and products, and the company's site describes it as an "End-to-End Automated AI Visual Inspection Platform" that is "No Code. Data & Vertical Agnostic." The reader it is written for is a quality, process or automation manager at a manufacturer who has inspection equipment and images of defects but no machine learning staff. The company lists an address in San Mateo, California. Its pitch is speed: "live in days" on the home page, with a counter reading "< 5 days to production deployment".

What you build

You build inspection models, not apps. The home page lays the workflow out as 5 steps in a row: Create, Upload, Label, Train, Deploy. You upload images, label the defects, train a custom model for your specific needs, and deploy it. Nothing in that flow asks for programming: the page says you can "train and deploy custom AI models without any programming skills". A company blog post about no-code computer vision repeats the idea, saying the platform lets non-technical users train and deploy high-performing models, with examples such as submicron defects in wafers, cracks in solar panels and anomalies in PCB solder joints.

The data requirement is deliberately small. The site says you can get "high-quality visual assessments using just 20-40 images per defect class", and the blog says you can build production-grade models with as few as 20 to 40 images per defect class. The home page also offers a try-it-now box, "Detect anomalies with a single image", where you upload one image.

The platform's model types are named on the home page under "Generate High-Accuracy AI Models For Defect Classification": Classification, Object Detection and Segmentation. Its solution areas break down like this.

• Defect Classification is described as automated defect classification with low false positives.
• Defect Detection is submicron defect detection with high accuracy, powered by advanced image processing.
• Defect Segmentation masks the exact defect area for higher accuracy.
• Defect Review is described as getting up and running with rapid defect review based on human input.
• Defect Monitoring categorizes and classifies defects based on custom input from the inspection team.
• Labeling Consistency is listed as improving dataset quality, enhancing model accuracy and ensuring reliable predictions.

The product menu also names Advanced Process Control and Virtual Metrology, which take the platform past image review into the process itself. The company blog describes Virtual Metrology as predicting process parameters without physical measurements, Advanced Process Control as feeding inspection insights back into production for yield optimization, and a predictive maintenance feature that flags anomalies for proactive equipment servicing. The site says the system includes "Continuous Learning": "AI that improves accuracy over time with active learning capabilities", and the blog adds that the models improve via active learning from new data.

Deployment is a choice between "secure on-premise or scalable cloud-based solutions", and the site says the AI integrates into your existing detection system either way. The FAQ answers the hardware question directly: Averroes works on existing visual inspection equipment, "whether it's KLA Tencor, AOI, or Onto equipment", and "No additional hardware purchase is needed". The blog describes the cloud and on-premise options as a way to meet IT and compliance needs, and says the platform is typically deployed per team or client rather than as a multi-tenant platform.

The company frames the product as a way of "Smartening Up Existing Capital Equipment": shorten the cycle, increase yield and improve quality with automated AI inspection software, instead of replacing the machines on the line. The blog post makes the same point from the user's side, saying quality teams can act faster and more accurately "without overhauling current equipment or workflows", and that the platform stands out for precision and simplicity. It names semiconductors, electronics, solar, food production and oil and gas as the industries where defect detection accuracy and operational efficiency are non-negotiable.

The industries listed are Oil and Gas, Food and Beverage, Pharma, Semiconductor, Electronics, Drone and Solar, and the product menu adds Photomask. The page describes the system as working "across different data sets and hardware". There is also a Developers section and resources including a blog, webinars, a whitepaper and a help center.

The evidence

The numbers are the company's own claims and read as marketing copy. The home page shows 99.97% defect detection, 10× faster than manual review and under 5 days to production deployment. A feature card promises "99%+ defect detection and 98.5%+ object detection rates with near-zero false positives". A closing block lists more than 99% classification accuracy, more than 98.5% object detection accuracy and more than 97.7% segmentation accuracy. The site calls itself "The #1 AI visual inspection platform", a claim the page makes about itself.

Two cost claims come with them. A feature card says automating inspections can "save 300+ hours monthly per application and boost product quality by 30%". A testimonial from an Automation Manager at a semiconductor OEM says: "We've experienced a remarkable 40-60% increase in submicron defect detection. Averroes has saved us over 300 hours/month/app of labor and increased our productivity by more than 30%." That is the only named customer voice on the page, and it is not tied to a company name.

The site publishes no prices; the calls to action are "Get a Personalized Demo" and "Request a Demo Now", and the company's own blog says the platform may require enterprise budgets or custom quotes. Budget a conversation, not a credit card. The same blog post lists its own caveats: initial data prep is still needed for new or niche defect types, and QA teams new to AI workflows face some learning curve. A Data Processing Agreement is linked from the footer next to the terms and privacy policy, and the site also runs a partner program.

Who it suits

Averroes fits a manufacturer that already inspects at volume with automated optical inspection or similar equipment and wants better classification of what that equipment finds. The 20 to 40 image figure and the on-premise option suit semiconductor, electronics and solar lines where images cannot leave the plant, and where defect classes are known and can be labeled by your own quality staff.

The caveats are straightforward. It is a vertical tool built for manufacturing inspection, so retail analytics, document capture or general object recognition are outside its scope. Pricing runs through a demo and a quote, so a small workshop cannot self-serve. And the accuracy figures are the vendor's headline numbers, so run a pilot on your own parts and your own defect classes before you rely on them.