Strike48 launches on-premises AI security appliance

Strike48 has launched an on-premises security operations appliance for AI agents, aimed at organisations that need to keep security data inside tightly controlled environments.

The appliance is available in on-premises and fully air-gapped configurations, allowing security teams to run autonomous security operations within infrastructure they control. AI models are loaded directly onto the appliance rather than accessed through externally hosted services.

The launch expands Strike48's deployment options beyond cloud and private cloud environments. The company now supports four deployment models: cloud, private cloud, on-premises, and fully air-gapped.

The broader range addresses a hurdle many companies face as they test AI for security work. Sensitive security telemetry may be prohibited from leaving a site or country, while some organisations remain cautious about how third-party AI models handle confidential data.

Strike48 is also positioning the appliance as a way to address cost concerns. Local inference can help customers avoid the uncertainty of token-based usage charges, which can rise as AI agent activity increases.

For customers with the strictest controls, the fully air-gapped version is designed to operate without a connection to external networks. The appliance can be supplied on dedicated hardware or deployed as a software appliance on infrastructure that meets Strike48's requirements.

Strike48, a product brand from Devo Technology, focuses on what it describes as agentic orchestration for security and IT environments. Its architecture combines cognitive and deterministic agents and is designed to work with different AI models rather than depend on a single model provider.

Regulated environments

The launch is aimed at sectors where data sovereignty and operational control are central concerns, including enterprises operating under strict regulatory requirements. In those settings, sending operational security data to cloud-based AI systems can limit adoption beyond trial projects.

Managed Security Service Providers may also see a use case for the appliance. Strike48 said it could help service providers offer AI-driven security operations to customers that do not want cloud-based AI deployments because of security, regulatory, data residency, or cost concerns.

Tim Leehealey, Co-Founder and President of Strike48, said the company was trying to match different customer operating environments rather than push a single deployment model.

"One of our core commitments is to meet enterprises where they are, and that means delivering the Strike48 platform in a way that matches their operational, security, and regulatory requirements," said Tim Leehealey, Co-Founder and President of Strike48.

"The path to agentic operations isn't going to look the same for every security and IT team. We're giving organizations the flexibility to adopt these capabilities on their terms," Leehealey said.

Local control

Strike48's argument rests on the idea that AI in security operations will be adopted unevenly, depending on how much control customers need over both data and infrastructure. Some large enterprises and public sector bodies have resisted relying on cloud services for functions tied to incident response, telemetry analysis, and security automation.

By moving models onto customer-controlled infrastructure, Strike48 is positioning the appliance as an option for organisations that want AI assistance in the security operations centre without moving underlying data into external systems. The approach also allows organisations to keep data on-site or within national borders.

Maidar Secure, cited by Strike48 as a partner user perspective, linked the product to concerns about retaining direct control over sensitive information.

"For many companies and institutions, retaining full control over sensitive security data is non-negotiable," said Timothy Whitaker, Engineering Team Lead and Lead Developer at Maidar Secure.

"The Strike48 appliance provides a practical way to deploy agentic security operations within infrastructure they control. As local AI models become more capable and efficient, organizations can scale these capabilities while maintaining data sovereignty, predictable costs, and operational resilience," Whitaker said.

The launch adds to a broader push across the cybersecurity market to offer AI tools in formats that fit existing compliance and data-handling rules, rather than requiring customers to move sensitive workloads to the cloud.