Ziroh Labs’ Kompact AI: Making High-Performance AI Possible Without GPUs

Could you tell us something about Ziroh Labs
Ziroh Labs started around six to seven years ago with a focus on encryption—specifically, computing securely on encrypted data using fully homomorphic encryption. The core team came from a mathematics and cryptography background, which naturally evolved into applying similar problem-solving approaches in AI. Over time, this foundation led us to build product-ready AI systems that integrate security, performance, and scalability—culminating in Kompact AI.

How is Kompact AI different from the AI systems most companies use today?
Most AI models, like ChatGPT, are closed and proprietary. However, the emergence of open-source models—beginning with Meta’s LLaMA—has fundamentally changed the landscape. Open models allow organizations and even countries to create domain-specific or culturally contextual AI. Kompact AI is designed around this philosophy. It helps businesses build, deploy, and optimize AI locally and efficiently, without depending on massive GPU infrastructure. The idea is to make AI both open and accessible.

Could you elaborate on how this technology will make AI more affordable for smaller companies?
Today, most AI workloads depend heavily on GPUs, which are expensive to procure and maintain. Kompact AI, on the other hand, runs efficiently on CPUs—the processors already found in most existing servers and devices. Advances from Intel, AMD, and Qualcomm now allow CPUs to handle complex computations once exclusive to GPUs. This means businesses can deploy powerful AI systems without huge capital investment. It significantly lowers barriers to entry while reducing energy consumption, making AI affordable, scalable, and sustainable for smaller enterprises and developing markets.

How do you think Kompact AI could help industries like education and healthcare in more practical ways?
AI can truly transform sectors like education and healthcare when it becomes accessible. In education, Kompact AI can power interactive tutors that assist students in multiple regional languages, providing instant feedback and personalized learning support. In healthcare, AI can help doctors record and analyze consultations, support diagnostic processes, and extend medical advice to underserved areas. In mental health, it can act as a first layer of support—helping people express and process emotions in a safe, responsive way. These are small but meaningful interventions that, at scale, can create social impact.

Ziroh Labs also collaborated with IIT Madras. What role did the partnership play in making Kompact AI possible?
Our collaboration with IIT Madras has been pivotal. Together, we established a Center for AI Research focused on sustainability, CPU-based performance, and security in AI systems. IITM’s academic validation and research ecosystem have helped refine our models and test them across varied use cases. The partnership also supports our work in Indian-language AI and cybersecurity applications, ensuring that our innovations align with India’s vision of self-reliant, energy-efficient technology.