However, the AI is uniquely enriched with analysis results generated by the Onekey Platform. Users can query security information and firmware analysis results in natural language and receive answers based on the platform's actual findings. Developed entirely by Onekey, this technology significantly improves the accuracy, transparency, and traceability of AI-generated responses.

"The goal in product cyber security isn't to generate answers that merely sound convincing. What matters is technical accuracy and reliability," said Jan Wendenburg, CEO of Onekey. "The Onekey AI Agent combines the power of modern AI with the deep analysis capabilities of our platform. Every response is grounded in verifiable security data." The AI operates directly on the platform's analysis results, including firmware extraction, binary inspection, Software Bills of Materials (SBOMs), component analysis, and vulnerability intelligence.

The new functionality is fully integrated into the Onekey platform. Following the current beta programme, the initial release will be available in September and will continuously expand. The goal is to help security teams move more quickly to transform highly complex firmware data into actionable insights and informed decisions.

Deterministic AI instead of generic responses

Generative AI has demonstrated the power of natural language interfaces. However, professional security environments require a higher standard: Responses must be transparent, verifiable, and based on concrete technical evidence.

"The market is currently focused on AI copilots, chatbots, and agentic AI. However, in cyber security, simply layering a language model on top of existing workflows is not enough," Wendenburg explained. "With the Onekey AI Agent, we've deliberately adopted an 'evidence first' rather than an 'AI first' approach. What matters isn't whether AI can generate a plausible answer—it is whether every answer is backed by verifiable technical facts." He continued: "Our AI Agent does not generate assumptions or generic responses from a general-purpose AI model. Every answer is based on the Onekey Platform's actual findings—from firmware extraction and binary inspection to SBOM detection, component analysis, vulnerability intelligence, and impact context. The result is responses that are transparent, contextual, and technically reliable."

"The second key differentiator is contextual awareness," Wendenburg added. "The Onekey AI Agent automatically recognises whether a user is analysing firmware, reviewing an SBOM, or assessing vulnerabilities. It understands the user's workflow and provides assistance exactly where security teams need it most."

The platform also supports the Bring Your Own Model (BYOM) and Bring Your Own Key (BYOK) approaches. These approaches allow organisations to tailor their AI deployments to their specific business, industry, and regulatory requirements.

Four-stage AI roadmap

The new Onekey AI Agent represents the first milestone in Onekey's four-stage AI roadmap. The company's vision is to transform the Onekey security platform into a comprehensive product security assistant that complements, rather than replaces, human decision-making.

In the second phase, organisations will be able to securely connect their AI agents, applications, and automated workflows to the platform. The third phase will see the AI evolve from an information system into an intelligent security assistant. The fourth stage will support complete product security workflows. Over time, the platform will assist with a broad range of operational product security tasks while maintaining full transparency and traceability.

"Our AI roadmap demonstrates how Decision Intelligence becomes reality step by step—as a trusted assistant for those who make security-critical decisions every day," said Wendenburg.