Lattica has exited stealth mode, introducing a groundbreaking cloud platform designed to tackle one of the most significant hurdles in artificial intelligence: data privacy. Powered by Fully Homomorphic Encryption (FHE), the startup's technology allows AI models to be queried using encrypted data – ensuring raw data remains unseen even by Lattica or the AI model provider. This innovation, valuable for both AI service providers seeking simpler compliance and end-users demanding confidentiality, is backed by $3.25 million in pre-seed funding led by Konstantin Lomashuk's Cyber Fund, with notable participation from Sandeep Nailwal, co-founder of Polygon Network and Sentient: The Open AGI Foundation, among others.
The rapid advancement of AI presents immense opportunities, but its reliance on vast datasets creates inherent privacy and security risks. These concerns are particularly acute in regulated industries like healthcare, finance, and government, often slowing or preventing AI adoption where it could provide substantial benefits. The Tel Aviv-based startup was founded to directly address this challenge, enabling organizations to harness the power of AI while maintaining the highest standards of data confidentiality.
FHE represents a paradigm shift in secure computation, allowing complex operations directly on encrypted data. With Lattica's platform, user data is encrypted locally before being sent for processing and only decrypted upon return, guaranteeing privacy throughout the interaction. While theoretically powerful, FHE adoption has historically faced challenges like high computational costs and implementation complexity. Lattica addresses these directly through its comprehensive "FHE-as-a-Service" platform, optimizing the entire stack from the underlying inference engine to the user client, making secure AI practical.
Central to Lattica's platform is its proprietary Homomorphic Encryption Abstraction Layer (HEAL). This technology layer acts as a universal translator, optimizing and standardizing FHE performance across diverse hardware environments – from GPUs and TPUs to CPUs and specialized accelerators like FPGAs and ASICs. HEAL ensures that secure AI computation via FHE can be as seamless and accessible as traditional cloud AI services.
"The potential of AI is undeniable, but so are the privacy concerns that hold it back, especially with sensitive data," said Dr. Rotem Tsabary, Founder and CEO of Lattica, who holds a PhD in lattice-based cryptography from the renowned Weizmann Institute of Science. "We recognized that by aligning breakthroughs in hardware acceleration with targeted software approaches, we could make FHE truly practical for neural networks and other AI models. Our mission is to unlock secure AI for everyone."
The $3.25 million pre-seed investment, led by Cyber Fund, underscores investor confidence in Lattica's deep-tech approach and its potential market impact. "Lattica is tackling a fundamental challenge at the intersection of AI and cryptography," commented Konstantin Lomashuk, Managing Partner at Cyber Fund. "Their work in making FHE practical is the kind of foundational innovation that paves the way for the next generation of secure AI applications."
"Lattica’s focus on delivering a scalable and practical FHE solution is set to transform how sensitive data is handled within the AI ecosystem, making secure AI a tangible reality," added Sandeep Nailwal.
Lattica is initially focusing on applications in healthcare and finance.
