Lattica, a Tel Aviv-based startup focused on encrypted AI computation, emerged from stealth mode on April 23 with a cloud-based platform that allows AI models to process encrypted data without decryption.
Notably, the company’s approach uses Fully Homomorphic Encryption (FHE), a cryptographic method long considered a breakthrough for privacy-preserving computation. Lattica is backed by Cyber Fund and Sandeep Nailwal, the co-founder of Polygon and Sentient. The company aims to address growing concerns about AI security.
Its focus is on regulated sectors such as healthcare, finance, and government.
Dr. Rotem Tsabary, Lattica’s founder and CEO, noted that:
By combining the advancements of hardware acceleration with software-based optimization, we realized that not only could we improve FHE efficiency to the point of commercial viability, but use it to solve critical data dilemmas holding back AI’s adoption in sensitive industries.“
Dr. Tsabary, who holds a PhD in lattice-based cryptography from the Weizmann Institute of Science, said Lattica’s system builds on the shared mathematical underpinnings between FHE and machine learning. Significantly, the result is a hardware-agnostic platform that enables encrypted AI operations. It supports GPUs, TPUs, CPUs, and custom hardware like ASICs and FPGAs.
How Lattica’s HEAL Layer Advances Secure AI Deployment
At the core of the platform is the Homomorphic Encryption Abstraction Layer (HEAL), a component that standardizes FHE acceleration across hardware types. According to the company, HEAL bridges the gap between encrypted applications and AI algorithms, allowing secure computation at scale.
Lattica’s emergence comes amid growing industry anxiety about AI-related data risks. Meanwhile, the company has also released demos of its platform alongside findings from a community survey on FHE. Among respondents, 71% agreed that future FHE adoption would depend on both software and hardware improvements — a premise reflected in Lattica’s design.
Sandeep Nailwal, who invested in the company, emphasized its practical orientation noting:
Lattica has made FHE a reality that is both practical and scalable. Their product-first approach transforms sensitive data processing in the AI ecosystem”
FHE allows computations to be performed on encrypted data, preserving privacy without sacrificing usability — a long-standing challenge due to the technique’s intensive computational requirements. Lattica claims its cloud-first approach helps overcome these limitations, making it suitable for industries that manage sensitive information.
Potential use cases include secure data analysis for medical research, encrypted financial transactions, and private identity verification systems. Lattica plans to expand its services as it seeks further adoption among enterprise clients.
