Ultra Security, an agentic security platform offering live oversight and controls for autonomous AI agents, has launched publicly after a year in stealth, backed by $10m in seed funding.
The round drew support from Ludlow Ventures, Hidden Capital, Entrée Capital, Detroit Venture Partners, firedrop and FDVC. Angel investors from the security and SaaS communities also took part.
Ultra positions itself as a checkpoint between AI agents and the MCP servers and connectors they rely on. Each tool call, data read and prompt request passes through the platform, allowing organisations to monitor and control agent behaviour while it happens rather than reviewing it later.
The company argues that agents have evolved from simple chatbots into software with genuine access to live systems, handling work across email, code and wider workflows. MCP is fast becoming the default route through which agents reach tools, APIs and sensitive data, yet the tooling designed to monitor it has failed to keep pace.
Web application firewalls and API gateways were designed to track traffic between people and systems, while cloud posture tools only assess configuration at a fixed moment.
According to Ultra, none of these can observe an agent’s actions within a live session, leaving many teams unable to say what their agents did, when, or which data they accessed. Ultra also anticipates a future where agents operate across all kinds of devices and hardware, making control over them essential.
The platform combines four core capabilities. Its guardrails convert plain-language policies into real-time enforcement on every tool call, blocking credential access, redacting PII, protecting PHI, limiting abuse and preventing data moving between connectors. These can apply to an entire organisation, a single team or one device.
Governance controls let firms specify which agents, connectors and tools each person or team may use. Observability features log and trace every agent-to-tool interaction, with AI-native anomaly detection raising alerts. A universal bridge provides connectivity across agents and connectors without per-client configuration, and the system is vendor- and model-agnostic.
Ultra was founded in 2025 by Chase Lee, previously founder and CEO of Trustpage, which Vanta acquired. Lee spotted mounting weaknesses in AI security as agents linked into tools, services and wider technology stacks that existing security products were never designed to handle.
During its stealth period, the team worked alongside customers and design partners to simplify agent security and AI governance for organisations regardless of size, security posture or AI maturity.
The company stresses that it was designed for agentic security from inception rather than adding AI features to an older product. It runs locally by default, keeping tool calls, payloads and audit trails on the user’s own machine, though a remote-hosted option exists.
With a founding team drawn from Vanta and Secureframe, Ultra already holds SOC 2 Type II, ISO 27001 and ISO 42001 certifications and is AARM Extended conformant. It aligns with NIST AI RMF and the Agentic Trust Controls, and has finished a FedRAMP 20x Class B audit with a 3PAO as it pursues FedRAMP certification.
Copyright © 2026 RegTech Analyst
