27 June 2025

Artificial intelligence is rapidly reshaping industrial quality control—and Italian multinational Antares Vision Group is leading the way. Together with Oròbix, its AI services affiliate, the company has launched AI-GO: a breakthrough AI platform designed to make vision inspection and traceability smarter, faster, and easier to deploy in real-world manufacturing environments.

AI-GO targets industries where accuracy, speed, and traceability are critical—such as pharmaceuticals, automotive, food & beverage, and agriculture. In an exclusive interview with MoveTheNeedle.news, Alessandro Baj Badino, Head of Corporate Communication & Investor Relations at Antares Vision Group, explained how this new AI platform transforms the way companies manage visual inspection.

From Concept to Shop Floor: The Real Gap AI-GO Fills

Baj Badino said that AI-GO was developed to address a clear gap: bringing AI into production in a scalable, reliable, and user-friendly way for visual inspection and traceability. “In manufacturing, pharmaceuticals and food & beverage, AI has shown great potential in detecting defects that traditional systems often miss. But the real challenge isn’t the algorithm’s accuracy: it’s the complexity of deploying and managing it on the shop floor. That’s exactly where AI-GO comes in: it enables the creation of models using just a few examples, no AI expertise required. It leverages pre-trained models and deep learning to reduce setup time, quickly adapt to new products, and reduce false rejects, ultimately improving overall quality.”

In other words, AI-GO is designed to bridge the gap between innovation and operational reality. The platform enables users to build robust AI models using just a few examples—no data science expertise required. It leverages deep learning and pre-trained models to reduce setup time, quickly adapt to new products, and cut down on false rejects. The result? Higher efficiency and improved quality outcomes.

Built for Real-World Manufacturing

Antares Vision Group has deep roots in highly regulated sectors, where AI must be not only accurate but explainable and robust under high-speed production conditions.

Key challenges AI-GO addresses include:

• Rare and hard-to-represent defects
• The need for continuous model updates and monitoring
• The requirement for a clear return on investment (ROI)
• Cultural and operational barriers to AI adoption on the factory floor

Technology That Works With You, Not Against You

At its core, AI-GO uses deep learning tailored for industrial computer vision. It supports classification, segmentation, and OCR, and can be trained with minimal data. “We provide a library of optimized pre-trained models,” said Manuela Bazzana, marketing & communication manager Oròbix, “and through fine-tuning we significantly reduce setup time and cost.”

AI-GO also integrates recent advances in AI, such as:

• Vision transformers
• Multimodal models
• Vision-Language Models (VLMs)

This enables human-like interactions: soon, users will be able to configure inspections by simply describing defects in natural language—no coding required.

Real Use Cases: Pharma, Glass, Agriculture, and More

In pharmaceutical settings, AI-GO is already proving its value through Smart Clearance, which automates and verifies line clearance during production changeovers. This use case:

• Accelerates changeovers
• Reduces error risks
• Enhances compliance through image-based traceability

Another successful case is the Glass Vial Inspection system, where AI-GO reduced false rejects caused by droplets and shadows. The model was trained solely on images of "good" samples using an anomaly detection approach—eliminating the need for time-consuming manual labelling.

AI-GO also powers diverse applications beyond pharma:

• Automotive: paint and weld defect detection
• Wood processing: surface measurement
• Logistics: label reading
• Agriculture: grape sorting and livestock behaviour monitoring

Depending on the customer’s needs, AI-GO can run on edge devices or in the cloud.

A Modular, Scalable AI Ecosystem

AI-GO also integrates seamlessly with Antares Vision Group’s DIAMIND suite. The company in fact claims that, when combined with DIAMIND suite, AI-GO becomes even more powerful. In agricultural projects, for instance, DIAMIND enables the collection and integration of heterogeneous data sources, from visual inspection to environmental and production parameters. This holistic approach allows users to uncover hidden trends and correlations that are not immediately visible through traditional analysis, supporting more informed decisions on quality, sustainability, and yield optimization.

Seamless Integration and Scalability

For clients already using Antares Vision inspection systems, AI-GO is natively integrated and requires no major infrastructure changes. It connects directly with existing sensors and vision systems, bringing deep learning capabilities without disruption.

The platform includes two main components:

• AI-GO Studio – a cloud-based training environment
• AI-GO Runtime – an edge module that runs models on-site, even offline

“Integration requires just five core functions,” noted Bazzana. “This is not a vague simplification, but a deliberate design choice to ensure true plug-and-play functionality. This means customers can begin generating value immediately, without the need for major changes to their infrastructure.”

Looking Ahead: From Inspection to Predictive Intelligence

What’s next for AI-GO? According to Badino, the platform is evolving toward becoming a full decision-support system. “We want to move beyond binary inspections,” he explained. “Our goal is to turn industrial vision into a tool for knowledge—not just detection.”

Antares Vision Group seems well positioned to take advantage of new opportunities in the computer vision market. The AI in computer vision market (not limited to industrial use) is projected to grow from $20+ billion in 2023 to over $50 billion by 2030, according to various market analysts. Within industrial applications, vision inspection is a major and fast-growing subset.

Despite the promise, adoption is still uneven due to:

• Complex integration with existing systems
• Lack of AI expertise on the factory floor
• Cultural resistance to new technologies
• Regulatory constraints, especially in life sciences
• Cost and ROI clarity, especially for small-to-midsize manufacturers

Solutions like AI-GO from Antares Vision Group specifically target these barriers by offering pre-built models, plug-and-play hardware integration, and intuitive UIs for non-experts.