Ultra Security has officially emerged from stealth with a US$10 million Seed round, unveiling an agentic security platform designed to protect autonomous runtimes. The Michigan-based company, founded in 2025 by former Trustpage CEO Chase Lee, addresses the rapid expansion of AI agents that now access real systems and sensitive data. Its launch comes as Model Context Protocol, or MCP, adoption accelerates faster than the security tooling built to monitor it.
Why the Timing Matters
Agents have evolved from simple chatbots into autonomous software with real access to email, code, and core business systems. MCP is rapidly becoming the standard way these agents connect to tools, APIs, and sensitive data, but adoption has outpaced monitoring capabilities. Ultra argues that existing security controls cannot see what an agent does inside a live session, leaving organizations without a clear record of actions, data access, or behavior.
How Ultra Secures Agentic Workflows
Ultra positions itself between AI agents and the MCP servers and connectors they call, giving organizations live visibility into every tool call, data read, and prompt request. Legacy tools such as web application firewalls and API gateways were built to monitor traffic between people and systems, not software that acts on its own. As a result, many teams cannot answer basic questions about what their agents are doing in real time, which is the problem Ultra was built to solve.
Guardrails, Governance, and Observability
The platform translates plain-language policies into enforcement on every tool call, allowing teams to block secret access, redact PII, protect PHI, and halt cross-connector data transfers. Governance controls let organizations decide exactly which agents, connectors, and tools each person or team can use. Ultra also logs and traces every agent-to-tool interaction, with AI-native anomaly detection providing alerts and a security backstop.
A Local-First Security Architecture
Ultra runs locally by default, keeping tool calls, payloads, and audit trails on the user's own machine, while a remote-hosted option is also available. The company emphasizes that it was purpose-built for autonomy rather than adding an AI feature to legacy security infrastructure. Its platform is vendor- and model-agnostic, serving as one universal bridge for agents and connectors without per-client setup.
Compliance and Market Readiness
Ultra enters the market with a strong compliance posture informed by its founding team's experience at Vanta and Secureframe. The company has already achieved SOC 2 Type II, ISO 27001, ISO 42001, and AARM Extended conformance, and it aligns with the NIST AI RMF and Agentic Trust Controls. Ultra has also completed a FedRAMP 20x Class B audit with a third-party assessment organization and is pursuing full FedRAMP certification.
Investor Backing and Availability
The US$10 million Seed round includes participation from Ludlow Ventures, Hidden Capital, Entrée Capital, Detroit Venture Partners, firedrop, and FDVC, along with angels from security and SaaS communities. Access to Ultra can be requested today at ultra.security. The company is headquartered in Birmingham, Michigan and led by Chase Lee, who previously founded Trustpage before its acquisition by Vanta.
Ultra Security's launch reflects a broader shift in cybersecurity as organizations move from securing human-driven workflows to governing autonomous software. By focusing on runtime visibility, policy enforcement, and compliance from day one, the company aims to give security leaders, engineers, and non-technical teams a unified way to manage AI agents. With the agentic ecosystem expanding quickly, Ultra's approach may help define how enterprises safely adopt autonomous runtimes.