A car body glides down the line, past a bank of cameras mounted on gantries overhead. In less than a tenth of a second, the system decides whether every panel, weld and label is exactly as it should be — catching flaws as small as 20 microns, finer than a human hair. Nothing slows down. Nothing slips past. That is the factory floor Frinks AI is building, one production line at a time.
Frinks AI is a Bengaluru deep-tech startup founded in 2020 by IIT Hyderabad alumni Aditya Agrawal, Dharmgya Sharma and Subhra S. Bhattacherjee. What began as a research project at IIT Hyderabad is now a Vision AI platform running on more than 1,000 production lines around the world, watching over cars, consumer goods, medical devices, cement and steel — and guaranteeing 99.99% inspection accuracy while it does.
The 15% problem
Machine vision has watched factory lines for 50 years, but the old rule-based systems could only handle a small slice of what a real line throws at them — the company says less than 15% of typical inspection scenarios. Every new part, every change in lighting or angle, meant new rules written by hand, and most defects still needed a human eye at the end of the line.
Frinks AI took a different route. The team built foundational vision models purpose-built for manufacturing — general-purpose AI systems that understand what a good part looks like across products and materials, then fine-tune on a small set of a customer’s own images for each specific job. The result, the company says, is inspection that runs in under 100 milliseconds, spots defects down to 20 microns, and carries a 99.99% accuracy guarantee.
What it watches
The platform covers the whole inspection job: assembly-line inspection, cosmetic defect detection, dimension measurement and label verification. Engineers set it up through a no-code platform — configure the workflow, train the model, deploy the inspection and monitor performance — without writing code or rebuilding the line around it.
It is hardware-agnostic by design, working with the cameras, PLCs, robotic arms, HMIs and servers a factory already has, and it goes beyond detection to real-time action: flagging a defect, triggering a rejection, and keeping a full traceable record of every part inspected.
More Read
The numbers factories care about
The company says its customers see 30% higher throughput from faster inspections, 60% lower cost of quality control, and 80% fewer product recalls. The platform is already in use across automotive, consumer goods, building materials, medical devices, cement and steel — sectors where a single missed defect can mean a recall, a warranty claim, or a lost customer.
The backing
In May 2025, Frinks AI raised $5.4 million in a pre-Series A round led by Prime Venture Partners, bringing its total funding to $6.25 million, with participation from Chiratae Ventures, Navam Capital and Ashok Atluri, founder of Zen Technologies. The company is backed by industry veterans including S Ramadorai, former MD and CEO of TCS, V Sumantran, former Executive Director at Tata Motors, Tarun Ramadorai, professor at Imperial College London, and Gopichand Katragadda, former Group CTO of Tata Sons, who support it as strategic investors and advisors. The startup has also been part of the Maruti Suzuki Innovation programme, and is using the fresh funding to scale globally, invest in research and development, and build its presence in the United States.
The factory floor has always been watched. Frinks AI’s bet is that it can finally be understood — every part, every line, every second, at a precision no human eye could ever match.
Keep reading: Two Minutes to Catch a Hacker: Inside BluSapphire, the Hyderabad Platform Guarding 500 Enterprises · Hunting the Hunters: Inside FalconFeeds, the Bengaluru Startup Watching 19,000 Cybercriminals in Real Time · More Industrial Automation & Technology stories
Like stories like this? Add Exceptional Technology as a preferred source on Google to see more of our reporting in your search results and Discover feed.
