Ultra Security raises $10m seed to police what AI agents do

Ultra Security raises $10m seed to police what AI agents do

21 October 2026: Ultra Security has come out of stealth with $10 million in seed funding for a platform that sits between AI agents and the systems they act on, checking every tool call, data read and prompt request against policy as it happens. Ludlow Ventures, Hidden Capital, Entrée Capital, Detroit Venture Partners, Firedrop and FDVC invested, alongside individual security executives including Shrav Mehta, Tim Sadler and Christina Cacioppo.

Not yet independently verified. The round rests on Ultra Security's own release, which names no lead investor and does not say how the money will be spent; its certifications are as the company states them. We will update this when it can be confirmed, and remove this note.

A checkpoint on every tool call

The company's argument is that the security stack was built for people and fixed configurations. Web application firewalls and API gateways watch traffic between users and systems, and cloud posture tools check settings at a point in time, but none of them sees what an agent does inside a live session. Ultra's answer is to route agent activity through its own layer.

It sells four parts. Guardrails turn plain-language policies into enforcement on each call: blocking access to secrets and credentials, redacting personal and health data, rate-limiting and stopping data from moving between connectors, scoped to an organisation, a team or a single device. Governance decides which agents, connectors and tools each person or team may use. Observability logs and traces every agent-to-tool interaction and flags anomalies. Connectivity is a single bridge between agents and connectors, which Ultra describes as vendor and model agnostic.

It runs on the user's own machine by default, keeping tool calls, payloads and audit trails there, with a hosted option for those who want one.

Compliance claims from a company in stealth

Chase Lee founded the company in 2025; he previously founded and ran Trustpage, a security trust-centre product. Ultra says that while still in stealth it became certified to SOC 2 Type II, ISO 27001 and ISO 42001, the AI management system standard, that it conforms to the AARM Extended specification and aligns with the NIST AI Risk Management Framework, and that it has completed a FedRAMP 20x Class B audit with a third-party assessor and is working towards FedRAMP authorisation. For a seed-stage vendor that list is unusual, and it is aimed at the regulated buyers who most need an audit trail of agent actions. Buyers should ask for the certificates and the FedRAMP status in writing rather than relying on the release.

The release is datelined Detroit, while the company's own description places its headquarters in Birmingham, Michigan. Access is by request at ultra.security.

A crowded early market

Agent security has drawn a wave of start-ups since agents gained the ability to call tools over MCP and similar protocols. Our AI security solutions category lists three products so far and our agentic AI software category 13; Ultra joins them as a security company first, which it presents as its difference from orchestration platforms that added a security tab.

Company profile on TrustList: Ultra Security

Sources

• Ultra Security: Ultra Security Raises $10M Seed to Secure Autonomous Agents, 21 October 2026
• TheSaaSNews: Ultra Security raises $10M seed, 22 October 2026