As organizations move from AI assistants toward autonomous agents that can plan and execute tasks, controlling what those systems are allowed to do is becoming a central enterprise technology challenge. Edvance International Holdings and Singapore-based CWG Innovations are developing RealRelay.ai, an agentic AI platform built around a security layer designed to test AI agents before and after changes.
Edvance International Holdings Limited and CWG Innovations have signed a three-year strategic cooperation agreement to commercialize RealRelay.ai, a unified agentic AI platform being developed by the two companies.
The platform is designed to give enterprises and institutions a single environment for creating, governing and deploying AI agents without requiring users to write code. More than 300 AI models are available through the platform, with tasks matched to what the companies describe as an appropriate model.
The most notable element is its AI security layer. A dedicated security agent evaluates an AI agent before deployment and performs another assessment when that agent is modified. The approach is intended to address a growing problem with agentic systems: unlike conventional chatbots, agents can potentially interact with data, applications and enterprise systems and take actions on a user’s behalf.
That makes permissions, validation and continuous monitoring increasingly important components of enterprise agent infrastructure.
Security Becomes Part of Agent Deployment
RealRelay.ai is being designed to operate across cloud, on-premises and fully offline environments. That flexibility could be particularly relevant to organizations handling sensitive information or operating under data-residency and security requirements.
Edvance brings more than two decades of cybersecurity experience to the project, while CWG contributes AI development and deployment experience across ASEAN.
Under the agreement, Edvance will commercialize the platform through its cybersecurity distribution and enterprise customer channels. CWG will operate and manage the platform from Singapore. Intellectual property associated with the platform’s cybersecurity AI agents will be owned exclusively by Edvance.
The commercial model is also significant. Rather than positioning agentic AI solely as a project-based service, the companies are targeting subscription revenue, potentially creating a recurring enterprise software model around secure agent deployment.
Education Provides the Initial Use Case
Education is the first target market. Planned capabilities include personalized learning and no-code agent creation for students and educators.
Cybersecurity is expected to follow, with use cases including security operations center triage and automated adversary simulation. Financial technology has also been identified as a subsequent vertical.
The platform has already been tested through APEX Bootcamp: RealRelay Edition, where students and young professionals without technical or AI backgrounds built working agents during a two-day program. According to the companies, the program was oversubscribed and every participating team produced a minimum viable product.
The timing aligns with Singapore’s broader focus on AI capability and governance. Singapore’s Infocomm Media Development Authority introduced a Model AI Governance Framework for Agentic AI in January 2026, while the Cyber Security Agency of Singapore has issued guidance addressing security considerations for agentic AI.
Those initiatives reflect a wider industry transition: enterprises are moving from asking what an AI model can generate toward determining what an autonomous system should be permitted to execute.
Market Landscape
The agentic AI market is increasingly shifting toward infrastructure for model orchestration, agent governance, identity, permissions, monitoring, evaluation and security.
Technology providers including Microsoft, Google, AWS and Salesforce are developing enterprise agent platforms, while cybersecurity companies are building controls around AI applications and autonomous workflows.
RealRelay.ai takes a different but increasingly important approach by putting an explicit security-testing layer around agents. The model is particularly relevant for organizations that need to deploy multiple agents while maintaining control over their behavior and access to enterprise resources.
The major technical challenge will be ensuring that agent evaluations remain effective as models, tools, prompts and workflows change. Continuous testing and policy enforcement therefore become as important as the underlying AI models.
Top Insights
- RealRelay.ai combines more than 300 AI models with no-code agent creation and deployment across cloud, on-premises and offline environments.
- A dedicated security agent evaluates AI agents before deployment and when they change, creating a continuous trust layer around autonomous workflows.
- Edvance will commercialize the platform while CWG operates it from Singapore under a three-year strategic cooperation agreement.
- Education is the first target vertical, followed by cybersecurity use cases and planned expansion into fintech.
- The platform reflects a broader shift toward agent governance, where permissions, monitoring and security become core components of enterprise AI infrastructure.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
