Security for AI agents.
Built by the people who need it.

Truvant is an AI agent security platform built by a CISO who couldn't find the tools to secure AI adoption across his own organization.

AI agents are connecting to MCP servers with no security visibility. Organizations don't know what tools these servers expose, what permissions they grant, or what commands their AI agents are actually executing.

We saw this problem firsthand — not from the outside looking in, but from the CISO's chair. When AI agents started showing up in our engineering workflows, we had no way to scan what they were installing, no way to enforce policy on what they could do, and no centralized view of what was happening across the team.

We built Truvant because we needed it ourselves. It's the tool we wish existed when we started adopting AI agents.

Truvant scans MCP server packages for vulnerabilities and secrets, enforces security policies on AI agent behavior, monitors for drift and supply-chain changes, and scores the trustworthiness of remote MCP endpoints. It's a complete security platform for the MCP ecosystem — from a single developer's machine to an enterprise fleet.

Leadership

Michael Chomicz

Michael Chomicz

Founder

Michael is the Chief Information Security Officer at Elisity, where he leads application security, vulnerability management, and compliance. He established an AI-first SOC, embedded secure-SDLC and DevSecOps practices, and hardened cloud and infrastructure across the organization.

A former engineering architect at Cisco, Michael brings deep expertise in cloud, networking, and automation. He holds CCIE #36817 and has spent his career at the intersection of infrastructure, security, and operations — from building service provider networks across three continents to leading enterprise security programs.

He founded Truvant to solve the problem he was facing as a CISO: how to adopt AI agents safely when nobody had built the security tooling yet.

CCIE #36817 CISO, Elisity Former Cisco Architect AI-First SOC DevSecOps
Michał Garcarz

Michał Garcarz

Co-Founder

Engineering leader with three decades in network security and cybersecurity. Full-stack foundation across application & database engineering, operating systems, virtualization, and enterprise networking. Named inventor on 28 U.S. patent filings spanning security, enterprise networking, application intelligence, and machine learning.

Spent 11 years at Cisco delivering managed security services; in the last four years, incubated next-gen security and AI/ML solutions powering multiple Cisco products.

Today he builds and scales cybersecurity startups, partnering with Fortune 100/500 enterprises on threat visibility, segmentation, and measurable risk reduction. He combines hands-on architecture with product and go-to-market execution — turning research into shipped products that generate revenue and improve security outcomes.

28 U.S. Patents Former Cisco, 11 Years AI/ML Security Enterprise Networking

Why we built this

No visibility

MCP servers expose tools and permissions that no existing security product understands. You can't secure what you can't see.

No enforcement

AI agents execute commands autonomously. Without policy gates, a misconfigured agent can ssh into production, force-push to main, or run privileged containers.

No trust model

Remote MCP endpoints are proliferating with no standard for evaluating their trustworthiness. TLS alone doesn't tell you if the tools are safe.

No audit trail

Security teams need to know what AI agents did, when, and why it was allowed or blocked. That data didn't exist before Truvant.

Want to talk?

We're always happy to hear from security teams navigating AI adoption.