Mumbai : Lytmus AI, a Bengaluru-based startup building AI mentors for competitive exam preparation, has raised a ₹5 crore pre-seed round led by Boundless Ventures, an early-stage fund backing AI-native companies. The company will use the capital to strengthen its AI capabilities, accelerate product development, and expand student acquisition, with an initial focus on the NEET segment.
Founded in 2024 by IIT Bombay batchmates Ajit Kumar and Praveen, Lytmus AI starts from a premise the edtech industry has largely agreed on for years: students preparing for exams like NEET need more guidance than recorded content can give them. What Lytmus has built differently is the mechanism to actually deliver that guidance at scale.
The platform runs on two layers working together. An infrastructure layer trained on the teaching patterns of top subject experts gives the AI mentors the ability to explain concepts the way an experienced teacher would. An application layer sits on top of it, building and carrying forward memory and context for each student: where they get stuck, how they tend to make mistakes, how fast they are moving through a topic. Together, this lets the mentor do more than answer a question in isolation. It can resolve a doubt in the moment, work out why a student keeps making the same kind of mistake, and proactively suggest the next step before the student loses momentum, in a way that responds to that specific student’s history rather than a generic script.
The company has assembled an academic team that includes Gaurav Gupta in Physics with over 12 years of teaching experience and a student following of 60,000, Dr. Gargi Singh in Biology with a following of over 70,000, and Hemant Umre in Chemistry with over five years of experience. The platform has been used by more than 16,000 students in the last 90 days, with students completing their daily practice up to three times more when guided by an AI mentor.
“In the last decade, edtech solved access to content. But it traded student-teacher interaction for scale, and exam outcomes did not follow. Students prepare alone, with no one to guide their next step. We built Lytmus to close that gap, not with more content, but with mentorship that knows each student and can actually get them to results. No student should have to prepare alone.” said, Ajit Kumar, Co-Founder and CEO, Lytmus AI
Ajit is a second-time entrepreneur and IIT Bombay alumnus with experience across product management, consulting, and startup operations. He previously built ventures in the agri-tech sector and led product initiatives across consumer and healthcare technology platforms. Co-founder and CTO Praveen bring over 10 years of experience in AI and machine learning and is also a second-time entrepreneur, having previously built ventures in the food-tech sector. Prior to Lytmus AI, he served as Staff Data Scientist at Simpl, where he led the data science team.
For Boundless Ventures, the investment reflects the fund’s conviction in AI-native companies solving the harder, less obvious half of the problem: not whether students need guidance, but how to engineer a product that can actually deliver it.
“What Lytmus has done is technically ambitious and commercially sharp at the same time. They have combined memory and context architecture with subject expertise from teachers who have genuinely earned student trust, and built a product that compounds the longer a student uses it. That is not a feature. That is a structural moat. In a market where two million students sit NEET every year and the vast majority go through it without real guidance, the team that cracks personalized mentorship at scale wins something significant.” said, Natasha Malpani, Founder and General Partner, Boundless Ventures
The investment comes at a time when education is moving beyond content delivery toward adaptive, relationship-driven learning tools. With NEET attracting well over two million registered candidates annually, the addressable opportunity for personalised academic support at the preparation stage is substantial and remains largely underserved by existing platforms.