Amid fast, complex production lines, how can manufacturers ensure consistent quality? Answering this, Kesava Prasad TD of VerifygnTech shares with EFY’s Nitisha Dubey how their AI-driven inspection systems—combining advanced software and hardware—offer a smarter alternative to traditional methods.
Q. Can you describe your offerings and the specific audience you aim to serve?
A. We specialise in AI-based software for automating quality inspection across manufacturing industries. Our primary customers include automotive original equipment manufacturers (OEMs), Tier 1 and Tier 2 suppliers. Tier 1 suppliers serve OEMs directly, while Tier 2 suppliers provide components to Tier 1s. Recently, we expanded into the SME (small and medium-sized enterprise) segment, supporting businesses that manufacture small sheet metal and press parts, including those supplying to companies in the US. This year, our key focus is on SMEs in the automotive supply chain, helping them enhance efficiency and quality through automation, regardless of their position in the tiered supply structure.
Q. What type of solution do you offer — is it hardware, software, or both?
A. As a deep-tech company, our comprehensive offering combines both hardware and software. The products utilise advanced technologies that extend beyond software, I must say. Although we are not a large-scale manufacturer, we do conduct in-house assembly of our machines and hardware components. This enables us to deliver turnkey systems to our customers, ensuring seamless integration and optimal performance.
Q. What kind of components does your system inspect?
A. Our system primarily focuses on inspecting PCBs (printed circuit boards) and electronic assemblies, ensuring high accuracy in quality control. For example, it can detect bent pins on PCBs, verify correct wire crimping in relay modules, and identify wrong or misplaced components such as capacitors or resistors. Beyond these, the solution is versatile enough to support general component inspection across a wide range of industries, adapting to various manufacturing environments and quality standards.
Q. Can you share some real-world examples of how your solution ensures quality and prevents critical errors in manufacturing?
A. Yes, these will illustrate the kind of solutions we provide. One of our customers is a leading truck and truck engine manufacturer. Their engines contain critical components like bearing shells, whose absence can cause complete failure if the engine is run. Our AI-based system is deployed across all their manufacturing lines and plants in India to ensure the presence of these bearing shells.
In another case, we work with a two-wheeler manufacturer where we use handheld scanners—not typical barcode scanners, but advanced vision systems—to verify label placements. These scanners ensure that labels, mainly for export orders, are correctly positioned. Like, if a label meant for the fuel tank is mistakenly placed on the rear side, it could lead to compliance issues. Our system prevents such errors.
A third example involves a foundry, where we inspect castings for surface defects, such as dents or scratches. These flaws can vary significantly in shape and size, making them difficult to detect manually. Our AI software accurately identifies them.
Q. Do you embed your software in machines, or do you supply customers with separate hardware and software systems?
A. We provide our systems along with dedicated testing stations. One approach involves a standalone testing station, where the customer places the part into our machine, and the system performs the inspection internally. Alternatively, our solution can be integrated into the customer’s existing production line. In this setup, as parts or engines move along the conveyor, our system inspects them at a specific point on the line. It instantly determines whether the part is acceptable or defective.
Q: What happens in the production system when a defect or issue is detected?
A. The system is designed in such a way that if any issue or defect occurs, the production line will stop. If it is integrated into the customer’s line, it immediately triggers an alert to notify them in real-time. This means the process cannot move forward until the problem is resolved. The person must address and clear the issue before proceeding to the next stage.
Q. How does your system integrate with existing production lines?
A. Integration is straightforward. Our control systems communicate directly with existing factory equipment like conveyors or assembly lines.
Q. What technologies do you use in your machine vision solutions?
A. One of our key solutions is the AI-based computer vision, which enables automated visual inspection with remarkable accuracy. We also leverage deep learning to detect complex defects that traditional methods may miss. In addition, robotics is integrated for efficient handling and movement of components during the inspection process.
Our technology stack utilises Generative AI (GenAI). This standout feature enables us to create synthetic defect data, particularly when real-world defect samples are limited, thereby enhancing the robustness and accuracy of our models. This approach enables us to achieve accuracy levels of up to 99.5 per cent.
Q. What role does deep learning play in defect detection?
A. Deep learning is a core part of our defect detection systems. It enables highly accurate identification of visual anomalies.
Q. How does real-time data analytics improve efficiency?
A. We are building predictive capabilities. For instance, consistent paint defects can signal issues in the painting process. By analysing such patterns in real-time, manufacturers can act before defects escalate.
Q. What are the key hardware/software components in your solution?
The key hardware components in our solution include a camera, lighting, a control panel, and an edge processing unit, which is a high-performance computer. The software components consist of AI/ML (artificial intelligence/machine learning) models, control interfaces, and integration tools, with all processing performed locally on-site rather than in the cloud.
Q. What is the exact cost if a customer wants to buy this product?
A. The cost of the product can vary depending on the level of automation and specific requirements. It generally ranges from ₹500,000 to ₹2 million. A basic system, which involves more manual effort, such as manually placing and checking each part, starts at around ₹500,000 to ₹700,000. On the other hand, fully automated systems, where parts are automatically placed, inspected, and sorted, tend to be more expensive, typically ranging from ₹1.5 to ₹2 million.
The pricing also depends on other factors, such as the complexity of the inspection, the number of parts being checked, and the amount of hardware involved. You see, if the system needs to inspect all six sides of a component, it will require more cameras and sensors, which adds to the cost. The final pricing is customised based on the customer’s needs and the extent of automation required.
Q. How does your solution differ from competitors in terms of coverage and problem complexity handling?
A. While we do have competitors in the market, what sets us apart is the fact that we offer a truly end-to-end solution. Most of our competitors typically provide only the core components, such as the camera, lighting, and perhaps some software. In contrast, we take complete ownership of the problem.
When a customer approaches us, we handle everything—from understanding how the part should be positioned and integrated into the production line, to designing and supplying all the mechanical, electrical, and control systems required. We fully deploy the system on-site, ensuring the customer does not have to worry about anything.
Another key differentiator is the complexity of the problems we solve. While many of our competitors focus on basic checks, such as verifying the presence of components in an assembled product, such as a mobile phone, we go much deeper. In addition to checking if the screen, camera, or audio port is present, we inspect for finer defects earlier in the process. This includes identifying missing components on a PCB, scratches, paint imperfections, incorrect labelling, and overall visual appearance. Our solutions are designed to catch issues that go beyond simple presence checks, offering more comprehensive quality assurance.
Q. Are you focused solely on design, or do you also specialise in product development?
A. We do not just design—we take the system from 3D design to manufacturing and assembly. That is our core strength. Each of our systems typically includes moving parts, mechanical components, electrical elements, electronics, and integrated lighting. The lighting aspect, in particular, is a critical part of our camera-based systems. We handle all lighting entirely in-house, customising it for each customer’s specific needs. We do not purchase off-the-shelf lighting—we design and manufacture it ourselves to ensure it meets the exact performance requirements of our systems.
Q. What is the current sourcing model for components in your system?
A. The primary imported components are the camera and lens, which are sourced from countries such as Singapore, Hong Kong, the US, and other global suppliers. Approximately 60 per cent of the materials are currently imported, while the remaining 40 per cent are developed in-house. Over time, we plan to reduce import dependency by identifying and working with Indian suppliers.
Q. Are there any major design challenges you face?
A. Design challenges are minimal. We have a strong in-house team for electronics, mechanical, and software design. The real challenge lies in sourcing, particularly in securing the right suppliers and ensuring the timely delivery of components.
Q. Can you describe your manufacturing facility, available machinery, and the size of your team?
A. Our main manufacturing facility is located in Pune, with additional activities taking place in our Thiruvananthapuram office. The Thiruvananthapuram branch primarily focuses on R&D (research and development) and software development. It is not a large-scale production setup, as most physical manufacturing tasks, such as fabrication or machining, are outsourced to vendors in the Pune region, which has a strong industrial ecosystem.
This approach helps us avoid the need for a sizeable in-house setup, especially since our business model does not involve high-volume production — we do not manufacture thousands of units per month. Instead, we operate on a low-volume, project-based model. As a startup, we currently handle around 25 to 30 projects per year, each involving custom solutions. Although we do not operate a full manufacturing line, we do conduct final assembly in-house. Our team comprises approximately 15 members.
Q. Are you planning to expand your manufacturing facilities?
A. Yes, we are considering expansion in South India, possibly near Bengaluru or Chennai, depending on customer demand.
Q. How many units have you sold so far?
A. We have completed over 50 installations across India. We are closing this fiscal year (March) with sales of approximately ₹17 million and aim to double that next year. Export orders could further increase our revenue.
Q. What ROI can manufacturers expect?
A. Typically, a 25–40% ROI (return on investment) through reduced labour, improved quality, and fewer recalls. For high-stakes industries such as aerospace, preventing one defect can save millions in potential losses or penalties.
Q. What are your revenue sources?
A. Our revenue comes from both hardware and software. We never sell hardware alone; the solutions are integrated with software.
Q. Are you seeking vendors or partners?
A. Yes, we are actively looking for strategic partners to support our growth. On one hand, we are seeking sales partners who can help us expand into new markets and reach a broader customer base. On the other hand, we are also seeking delivery partners who can oversee the deployment and support of our solutions in various regions, both within India and internationally. This approach allows us to scale efficiently without significantly increasing our internal team size, while ensuring quality implementation and customer support.
Q. What are the main challenges in implementing AI-powered vision systems?
A. Customer awareness is the biggest hurdle. Many expect immediate results, but AI systems require diverse and quality data over time. Lighting conditions, product variety, and other environmental factors impact system performance. Customers should understand that optimal accuracy, typically achieved at around 99 per cent, requires one to two months of refinement.
Q. What are your investment and growth plans for the next two years?
A. Our focus areas include team expansion, tapping into export markets, and scaling our existing products for mass deployment. With adequate funding, we aim to achieve a revenue of ₹150–200 million within the next two years.
Q. What are your plans for expanding in AI and automation?
A. We aim to serve more SMEs and MSMEs (micro, small and medium-sized enterprises), focusing on complex, small-batch production challenges. We are also combining inspection with measurement in single solutions, expanding our value offering.
