Two Silicon Valley companies just announced a partnership that tackles the biggest barrier to deploying AI safely at the edge: how to shrink powerful models small enough to run locally while keeping them secure from attacks. 1stProtect, a runtime security firm founded by veterans of CrowdStrike, Symantec, and Cisco, has partnered with Multiverse Computing, a leader in quantum-inspired AI model compression, to deliver integrated on-device AI inference with built-in protection.

What Problem Are These Companies Solving?

As enterprises race to deploy generative AI and autonomous AI agents into production, two major obstacles have emerged: cost and control. Cloud-based AI inference is expensive and creates data-sovereignty and latency challenges for regulated industries like healthcare, finance, and defense. Autonomous AI agents introduce new security risks that traditional security tools were never designed to handle. This partnership addresses both problems simultaneously by enabling AI to run entirely on-device, with no data sent to the cloud.

The integrated solution is purpose-built for environments where AI has been hardest to deploy safely, including air-gapped systems, operational-technology (OT) infrastructure, defense and critical infrastructure, healthcare, and financial services.

How Does the Technology Work?

The partnership combines two complementary technologies. Multiverse's CompactifAI compression technology reduces the size of leading large language models (LLMs), which are AI systems trained on vast amounts of text, by up to 95 percent with only 2 to 3 percent precision loss. This dramatic shrinkage enables roughly 4 to 12 times faster processing and 50 to 80 percent lower inference costs on hardware ranging from smartphones to industrial edge devices.

Once models are compressed small enough to run locally, 1stProtect's runtime enforcement layer wraps that on-device inference in continuous protection that responds in under 100 milliseconds. This protection blocks prompt injection attacks, model theft, and unauthorized agent actions the moment they occur, whether the device is online or completely offline.

Steps to Deploy Secure On-Device AI

• Model Compression: Use quantum-inspired compression techniques to reduce large language model size by up to 95 percent, enabling deployment on edge devices without sacrificing accuracy.
• Runtime Protection Layer: Wrap compressed models with real-time enforcement that monitors and blocks malicious actions, prompt injections, and unauthorized agent behavior in under 100 milliseconds.
• Offline Capability: Deploy the integrated solution on air-gapped systems and operational-technology infrastructure where cloud connectivity is unavailable or undesirable for security reasons.

"This partnership brings a uniquely secure inference solution to the edge for the benefit of our customers. As enterprises move into an era of agentic and AI-driven threats, they need the freedom to run powerful models anywhere without surrendering control," said Kervin Pillay, Chief Executive Officer at 1stProtect.

Kervin Pillay, Chief Executive Officer, 1stProtect

The combined offering builds on 1stProtect's AgentProtect, which is generally available today. Prospective customers can request an Agent Exposure Assessment or contact the sales team directly to learn more about deploying secure, compressed AI at the edge.

Why Does This Partnership Matter Now?

The timing reflects a broader shift in enterprise AI strategy. As organizations deploy autonomous AI agents that can take actions without human approval, the security risks multiply. Traditional endpoint detection tools and prompt filters were designed for a different era and cannot stop runtime threats from autonomous agents. By pairing model compression with runtime security, this partnership creates a complete solution that addresses both efficiency and safety concerns simultaneously.

"Compressing world-leading models so they fit on-device was the first breakthrough; making sure those agents run securely wherever they are deployed is what turns edge AI into an enterprise reality," noted Enrique Lizaso Olmos, Chief Executive Officer at Multiverse Computing.

Enrique Lizaso Olmos, Chief Executive Officer, Multiverse Computing

The partnership is particularly significant for industries with strict data-sovereignty requirements, unreliable connectivity, or latency-sensitive applications. Defense contractors, hospitals, financial institutions, and manufacturers can now deploy advanced AI capabilities without relying on cloud infrastructure or exposing sensitive data to external networks.