Kompact AI, a CPU-first platform from Ziroh Labs, launches to power high-volume enterprise AI workloads on readily available CPU infrastructure, targeting sectors like healthcare, education, retail, fintech, cybersecurity and enterprise automation.

Early partners are evaluating Kompact AI for AI-powered software testing, intelligent enterprise platforms, and sector-specific workflows, reducing the need to build AI stacks from scratch.

The platform prioritizes AI inference on CPUs to cut GPU dependency, enabling deployment close to existing applications and data, with a focus on cost control, performance, and data locality.

Industry and academic leaders, including IIIT-Bangalore professor S. Sadagopan, praised Kompact AI for its potential to boost India's deep-tech ecosystem with a frugal, energy-efficient AI approach.

Industry voices, led by Prof. Sadagopan, described Kompact AI as timely and energy-efficient, with the potential to spur a large ecosystem of partners, papers, and models in India’s AI landscape.

At the Bengaluru launch, academic and industry figures celebrated Kompact AI’s ecosystem and its promise of accessible, energy-efficient AI development and deployment.

Renowned cryptographer Whitfield Diffie endorses Kompact AI’s CPU-distribution approach, arguing it enhances accessibility and decentralization of AI capabilities.

The Bengaluru launch featured Diffie, underscoring decentralization and accessibility as core AI-infrastructure principles and praising CPU-based AI on widespread hardware.

Diffie’s high-profile endorsement positions CPU-based AI as a transformative way to broaden AI reach across common computing resources.

CEO Hrishikesh Dewan emphasizes making AI infrastructure accessible to bring AI closer to where businesses operate, improving cost control and performance.

Co-founder and CEO Dewan reiterates that accessibility, cost management, and data locality are central to enabling AI near primary business operations.

Ziroh Labs partners with IIT Guwahati on a Math Lab initiative to explore mathematical structures in large language and state-space models, plus a free 40-plus-hour AI course and separate certification.