Detect AI

Zenjin Detect AI

Find the models and agents that were already shipping.

Zenjin Detect AI finds unsanctioned models, agents, and MCP servers in your codebase, promotes them into a governed inventory, and produces AIBOM / AgBOM so you know what is actually running.

Request a demoAll solutions

At a glance

  • Scan

    GitHub org, day zero

  • Inventory

    Apps, models, MCP

  • AIBOM

    What the app runs

  • AgBOM

    What the agent can call

Find it, name it, put it in the catalog

Shadow models and MCP servers do not wait for a spreadsheet. Detect AI scans the org, promotes what it finds, and writes AIBOM / AgBOM for what actually runs.

Shadow AI discovery

Find unsanctioned LLMs, agents, and MCP servers in the codebase.

Live inventory

Models, applications, and MCP/agents with owners and lifecycle, not a spreadsheet.

AIBOM

Generate an AI bill of materials for applications headed to production.

AgBOM

Agent/MCP bill of materials plus BOM policies before something ships.

Dark metallic sphere found in black space, marked by thin scan lines
Request a demo

Today vs Zenjin

If it is not in inventory, it is a rumor.

Today

Shadow models with no owner.

Shadow AI lives in GitHub orgs, side projects, and vendor tools. Ownership is unclear until an auditor asks.

  • Side project
  • Unknown
  • Vendor tool
  • No owner
Unnamed shadow shards on a black field

With Zenjin

Applications, models, and MCP in one catalog.

Org scans feed Application, Model, and MCP/Agent inventory. AIBOM and AgBOM make the bill of materials part of the same record.

  • Scanned
  • Promoted
  • Owned
  • BOM
A violet catalog ring holding Application, Model, and MCP on gold collars

Scan, promote, bill of materials

Scan the org, name what you found, put a bill of materials on it.

  1. 01

    Scan the org

    GitHub organization scan, standard or deep, on day zero and on a cadence.

    A scan ring around unnamed repos

  2. 02

    Promote into inventory

    Repos become applications. Scan assets become models, MCPs, and agents with owners.

    Scan shards becoming Application, Model, and Agent on gold collars

  3. 03

    Generate BOMs

    AIBOM for applications, AgBOM for MCP/agents, with production policy checks.

    AIBOM and AgBOM glass packets with engraved gold bands

Who it is for

  • Platform
  • AppSec
  • ML ops
  • GRC

FAQ

  • Detect AI finds what exists. Inventory owns it. AIBOM / AgBOM describe the running system so governance is not a document someone wrote once.

  • GitHub org scan is the day-zero path. Promoted assets become the catalog the rest of the platform (Data Security Posture Management, agentic red teaming, evals) operates on.

  • AgBOM is the bill of materials for agents and MCP servers: tools, models, and components the agent can call, with production checks.

  • The org scan finds MCP usage. You promote those assets into MCP / agent inventory, then generate AgBOM for what is allowed to run.

  • Yes. Shadow models, agents, and MCP servers in the org are the point of the scan, including standard and deep GitHub organization scans.

  • You promote into inventory, attach owners, generate AIBOM / AgBOM, and the other products (Data Security Posture Management, evals, red teaming) operate on that catalog.

Related