Kusho AI Wants to Catch Software Bugs Before They Become a Problem

Writing software has become faster with AI, but keeping that software running smoothly is still a headache for engineering teams. Bengaluru-based Kusho AI is betting that AI can help there too. The startup is building agents that can test APIs, websites and mobile apps automatically, spotting problems before new code reaches users.

Founded in 2023 by Abhishek Saikia and Sourabh Gawande, Kusho AI takes on the repetitive work that usually comes with software testing. Its platform studies how an application works, figures out which scenarios need to be checked, creates the tests and runs them on its own. As the software changes, the system can also update those tests instead of making teams rewrite them every time.

The idea came partly from Saikia’s time at Flipkart, where he saw how much effort engineering teams spent making sure new releases didn’t break something that was already working. Kusho now acts almost like another checkpoint in the development process. It can test code before it is merged, flag risky releases and continue watching for issues even after the software has gone live.

That approach appears to be finding takers. Kusho says it works with more than 5,200 customers, with names such as Paytm, Ola Electric, Razorpay, Ultrahuman, HCL, Michelin and HP on the list. Its free product is also being used by around 35,000 developers and quality assurance professionals, while the platform has generated and run close to 10 million tests so far.

Kusho is now looking beyond early adoption. The startup says its annual recurring revenue is already in the late six figures in dollars and is growing month-on-month. It has raised $600,000 in pre-seed funding and plans to raise another round as it pushes deeper into banking, logistics and healthtech. While its team currently operates from Bengaluru, a large share of its customers are already in the US — giving Kusho a fairly global ambition for a seven-person startup.

For Kusho, the larger bet is simple: developers shouldn’t have to spend so much of their time checking whether yesterday’s software still works after today’s update. If AI can take over more of that maintenance work, engineering teams get to spend more time doing what they were hired for in the first place — building.