Gluware's Titan platform coordinates multiple AI agents — from observability platforms, service management tools and security systems — that want to make simultaneous network modifications.
When Jeff Gray and his co-founders showed their first network automation prototype to a major network equipment manufacturer over a decade ago, the response was dismissive. The vendor questioned why anyone would want a system that constantly modified network configurations, calling it something that would get “really really busy.” The prevailing wisdom held that networks should be set-it-and-forget-it, with changes avoided to prevent cascading outages.
The modern networking landscape sees things very differently. The threat landscape has fundamentally altered the industry’s approach to network changes. Nation-state actors, vulnerability exploits and evolving attack vectors now force enterprises to constantly patch, remediate and tune their networks.
Gray’s company was founded as Glue Networks and later evolved into Gluware. The company began with Cisco-only WAN automation before expanding into multi-vendor networking support that progressed from human-approved remediation to fully automated self-operating networks. Early customers required manual approval for each automated change. Today’s deployments automatically remediate configuration drift the moment it’s detected.
With the rise of AI, the industry now faces a new challenge: coordinating multiple AI agents that want to make simultaneous network modifications. Observability platforms, service management tools and security systems are all deploying their own AI agents. Without an arbiter, conflicting changes could trigger the very cascading outages that made network teams cautious about automation in the first place. Gluware is addressing this coordination problem with the launch of Titan.
“If all these companies are building their own agents, which agent is in charge?” Gray told Network World. “What we’re saying is that with Titan, it’s a validated framework. You’re able to fully abstract the network and trust it with Gluware and DIAL technology, and we can be the arbiter of what is safe and predictable.”
From configuration management to AI agent coordination
Gluware’s automation capabilities evolved through three distinct operational phases that built the foundation for Titan’s multi-agent coordination.
The first phase focused on configuration management and drift detection. Gluware’s system identified when network devices deviated from approved configurations and proposed fixes, but network operations teams manually reviewed and approved each remediation.
The second phase introduced automatic remediation. As customers gained confidence, they allowed the system to automatically correct devices that drifted from approved standards without human approval for each action. Current deployments represent what the industry calls self-operating networks, where configuration drift triggers instant automated correction.
Titan represents the third phase: system-determined operations where AI identifies new changes needed and executes them within defined risk parameters. This differs fundamentally from drift remediation, which restores known-good states. Titan handles new modifications from multiple AI systems that may conflict with each other. The platform coordinates between observability platforms, service management tools and security systems, each deploying AI agents that want to make simultaneous network changes. The validation framework extends from preventing configuration drift to preventing conflicts between competing AI agents.
Titan’s architecture for network and agentic AI coordination
Titan consists of a series of integrated components working together to solve the multi-agent coordination problem.
The Intelligent MCP Server uses the Model Context Protocol to coordinate between Gluware’s automation capabilities and external AI agents. The Gluware Agent executes the actual automation work between the MCP server and network devices. The Co-Pilot provides a natural language interface for network operations teams.
The MCP Server implements what Gluware calls a validation engine. Every action passes through verification before execution. This architecture allows third-party agents from observability platforms or service management tools to request network changes while Gluware maintains control over execution. Initial MCP service integrations include NetBox and ServiceNow.
Validated automation built on DIAL foundation
At the foundation of the automation is Gluware’s Device Interaction Automation Layer (DIAL), representing over one million developer hours across more than 300 releases. DIAL enables bidirectional communication with physical and virtual network devices across vendors and platforms, handling device-level complexity at scale.
The validation framework builds on the trust Gluware established through a decade of customer deployments. The same logic that convinced network teams to move from human-approved changes to automatic remediation now extends to coordinating multiple AI agents. Each proposed change passes through verification before execution, regardless of which AI system initiated the request.
“DIAL is underneath, Titan is on top, and then it’s a trusted framework for other third-party MCP servers to plug into,” Gray said.
