– By Mithran Sri Ram A.N.

Artificial Intelligence (AI) in production is the ability of computers to think and behave like humans, including recognizing and reacting to external and internal stimuli and even forecasting future occurrences. The machinery can identify when a tool is worn out, when something unexpected occurs, or even when something anticipated occurs, and then respond appropriately. AI has emerged as a logical consequence of the increasing capability of computers to do tasks that people have traditionally performed for themselves. AI’s strength lies in its ability to free up people’s imaginations to pursue other goals.

However, this technology does not replace humans; the best applications assist people in accomplishing what they are particularly excellent at. SwitchOn Co-Founder Mr. Aniruddha Banerjee, shares his views with CMR on how AI’s role in their success journey of 5 + years.

Q1. Can you share how the journey started for “Switch On”?

Ans. SwitchOn was founded in 2017 by Aniruddha (ex-Nvidia, ex-Samsung) and Avra (ex-Schneider, ex-Team Indus) with the vision of Eliminating Defects in Manufacturing.

Q2. How do you help manufacturing industries?

Ans. SwitchOn provides cutting-edge solutions to precision manufacturing industries, helping them to significantly decrease manufacturing defects through the implementation of a unique field-trainable AI system. The company’s unique approach to AI enables it to train deep-learning models with <200 images thereby drastically reducing the training time to deliver results within days while others take months to set up in the industry.

SwitchOn helps precision manufacturing Industries drastically reduce their manufacturing defects through the use of a one-of-a-kind field-trainable AI system. Automotive and CPG companies globally employ 25%+ of their workforce on inspecting product quality and yet face 500+ customer complaints per year per site leading to $120M+ in losses. The SwitchOn DeepInspect system can be trained by the Quality Inspectors on the field to inspect complex surface defects and eliminate customer complaints through AI-on-the-Edge. Leading global manufacturers such as SKF, ITC, Unilever, P&G, and others are using the SwitchOn platform to give quality superpowers to their Inspectors and ensure zero defects in their plants.

Q3. Can you share about services?

Ans. SwitchOn helps precision manufacturing Industries drastically reduce their manufacturing defects through the use of a one-of-a-kind field-trainable AI system. Automotive and CPG companies globally employ 25%+ of their workforce on inspecting product quality and yet face 500+ customer complaints per year per site leading to $120M+ in losses.

The SwitchOn DeepInspect system can be trained by the Quality Inspectors on the field to inspect complex surface defects and eliminate customer complaints through AI-on-the-Edge. Leading global manufacturers such as SKF, ITC, Unilever, P&G, and others are using the SwitchOn platform to give quality superpowers to their Inspectors and ensure zero defects in their plants.

Q4. Is AI an inevitable thing in any industry?  How far is it true?

Ans. The progress of AI in almost every field is being felt across the globe. However, the rise of models that are closer to Artificial General Intelligence like ChatGPT, and others have brought the creativity of AI to the forefront. While earlier AI was being used to automate majorly repetitive tasks such as recognizing handwriting, now it is being used for complex tasks like Identifying defects in products. The future is going to be even more interesting in terms of AI being used for creative writing and other things.  With this in mind, I believe that the adoption of AI is inevitable in any Industry, and it will be used to augment the capabilities, and competencies of humans.

Q5. How does “Switch On” help reduce manufacturing defects?

Ans. Even in modern manufacturing, while most of the manufacturing process is completely automated, the quality inspection and quality assurance processes are predominantly manual. Even on lines that manufacture products at 300+ products-per-minute, there is online quality personnel that still goes to the line and identifies defects. This is slow, error-prone, and leads to millions of defective products being shipped to customers from a single plant regularly. With the use of AI, we are able to do a 100% inspection of the manufactured products and reject the faulty products directly on the line to eliminate defects.

Q6. What is the customer satisfaction ratio for your products & services?

Ans. We are working with Fortune 100, Fortune 500 Enterprise customers, where we have started on a single site and then expanded across multiple sites. All the customers who have used our products, continue to use our products over multiple years and inspect ~30,00,000 products per month through a single system.

Q7. “Switchon” serves in multiple verticals/industries.  Can you elaborate in detail?

Ans. SwitchOn works in majorly the automotive and consumer-goods industries, where we work to inspect a myriad of products. In the automotive Industry, we are working with enterprise customers such as SKF, and others who are using our products to inspect critical bearings, engine parts, brake parts, and others. We, also, are working with large enterprise customers in the consumer-goods segment such as Unilever, P&G, ITC, and others who are using our products to inspect anything from glass/plastic bottles, soap bottles, beverages, caps and other products at speeds of 2000+ parts-per-minute.

Q8. What are the challenges “Switch On” face both in the market & with customers?

Ans. The major challenge we are facing is reaching customers in the manufacturing Industry due to the legacy nature of such Industries, and having the platform to be able to show the power of AI in their manufacturing environment. We are also looking to grow our product team to service the increasing demand from our customers.

Q9. Can you share about New Age Quality Inspection – available currently?

Ans. Most of the current generation of inspection systems are conventional systems. While they are good at detecting basic defects such as color, and presence-absence, but complex surface defects such as dents, and scratches, and a comprehensive inspection is still not possible. Hence, you need to still have a manual inspection even after installing the system which defeats the purpose. Also, the current generation of AI systems are extremely data-intensive and require 2+ months to set up, which is usually impractical for the typical manufacturing environment.

Q10. Can you share about the Deep Inspect System?

Ans. The SwitchOn DeepInspect system is a state-of-the-art AI system that has been built with the ability to train with <200 good images.

The system consists of three parts:

1. Training Platform: Our customers are able to upload <200 images to our platform and train AI models through our platform.
2. Inference Platform: Our customers can deploy the AI models trained through our training platform on the Edge, and inspect as well as reject products at 2000+ parts-per-minute.
3. Analytics Platform: Our customers can use our Analytics platform to analyze overall defect trends and automate the complete quality reporting. This can also be integrated with other third-party software to provide higher-level analytics to the upper-management and central teams.

Q11. Can you update us on your client Portfolio, and product uniqueness?

Ans. In the automotive Industry, we are working with enterprise customers such as SKF, and others who are using our products to inspect critical bearings, engine parts, brake parts, and others. We, also, are working with large enterprise customers in the consumer-goods segment such as Unilever, P&G, ITC, and others who are using our products to inspect anything from glass/plastic bottles, soap bottles, beverages, caps and other products at speeds of 2000+ parts-per-minute.

The SwitchOn DeepInspect system is a state-of-the-art AI system that has been built with the ability to train with <200 good images. The ability to get 99%+ accuracy without the need for defects means that we are able to deploy a product within 8 days and ensure that our customers are realizing value within less than a week of the deployment. The unprecedented accuracy within such a short time period makes the adoption of our customers uniquely fulfilling.

It’s becoming more and more about teamwork between people and robots.  Most industrial robots need a lot of human oversight, despite the widespread belief that they are self-sufficient and smart.  However, advancements in AI are making them smarter, leading to human-robot teams that are both safer and more productive.