Shadow AI Governance

Every AI tool on
your endpoints. Under control.

Kuntur discovers unauthorized AI tools across your fleet, lets you mark each one as approved or unauthorized, and gives you the audit trail compliance teams actually need.

Request early access See how it works

Cloud or self-hosted · Windows, macOS & Linux · Metadata only

kuntur-agent — endpoint scan
$ kuntur-agent scan --pretty
→ scanning apps, processes, IDE & browser extensions, MCP servers…

✓ Microsoft Copilot — approved
⚠ ChatGPT Desktop — FINDING: unauthorized AI client
✗ Cursor — FINDING: unapproved AI IDE
✓ GitHub Copilot (VS Code) — approved
⚠ Perplexity browser ext. — FINDING: unreviewed data access

findings: 3 open | checked: 147 processes | elapsed: 1.2s
35+
AI apps, IDEs and CLIs
detected out of the box
3
desktop platforms:
Windows, macOS, Linux
0
prompts, documents or
file contents collected
1
silent MSI for Intune or GPO
— agents enroll themselves
Features

Everything you need to govern
AI across your organization

A lightweight agent, a detection catalog and a governance dashboard — as a managed service or on your own servers.

Endpoint Discovery

The Kuntur agent continuously inventories installed AI apps, running processes, IDE and browser extensions, MCP servers and local models to surface every AI tool in use — approved or not.

Static Go agent · No dependencies

Detection as Code

Mark each AI tool as approved or unauthorized from the dashboard. Detection itself is a YAML catalog with embedded tests, so new tools can be added without touching the agent.

YAML detection rules · Dashboard policies

Audit Trail

Every triage action and policy change is logged with timestamp and actor. Export for SOC 2, ISO 27001, or any compliance framework your auditors require.

CSV export · Actor and timestamp

Findings Triage

Review findings one-by-one or bulk-update dozens at once. Mark as acknowledged, resolved, or suppressed. Your security team stays focused on what matters.

Bulk actions · Status tracking

Risk Dashboard

A real-time overview of your organization's Shadow AI posture: critical findings, affected endpoints, compliance status, and trends over time.

Real-time · No external deps

Suppression Rules

Permanently silence known-safe configurations without cluttering your findings. Apply suppressions by rule, endpoint tag, or specific item — with an audit entry for each.

Context-aware · Audited
How it works

Up and running in minutes

Use our managed service or run the server yourself. Roll agents out to endpoints at your own pace.

Step 01

Get your server

Start on our managed service, or run the server on your own infrastructure as a single container with SQLite storage: no external database, and endpoint data stays in your network.

docker compose up -d
# or use your tenant on
https://app.kuntur.io
Step 02

Decide what is allowed

Mark each AI tool as approved or unauthorized, and silence known-safe setups with suppression rules. Detection is driven by a YAML catalog you can extend.

id: chatgpt-desktop
category: desktop-app
procs:
- name: chatgpt
Step 03

Roll out agents

Deploy the lightweight agent via your existing endpoint management tool (Intune, GPO, Jamf or Kandji, Ansible). Each agent registers automatically and starts scanning.

# MDM / Intune script
./kuntur-agent install \
--server https://app.kuntur.io \
--key $ENROLLMENT_KEY
Compliance

Built for regulated environments

Kuntur generates the evidence your auditors need, and you choose where the data lives.

  • SOC 2 evidence
    Audit log of triage actions and policy changes with actor and timestamp. Findings export to CSV on demand.
  • ISO 27001 controls
    Supports A.8.19 (software installation) and A.8.32 (change management) with continuous monitoring.
  • Your choice of data residency
    Self-host the server in your own VPC or on-premises and endpoint data never touches Kuntur's infrastructure.
  • Data minimization by design
    The agent reports metadata only. Credentials are redacted on the endpoint, and browser history is reduced to visit counts for known AI services.
Sample findings export
ChatGPT Desktop · LAPTOP-04
no-unauthorized-llm · critical
Resolved
Cursor · DEV-12
unapproved-ai-ide · critical
Open
Perplexity Extension · LAPTOP-07
unreviewed-browser-ext · warning
Acknowledged
Gemini CLI · DEV-01
no-unauthorized-llm · serious
Suppressed
Claude Desktop · LAPTOP-22
no-unauthorized-llm · serious
Open
Early access

Be first to know
when Kuntur ships

We're onboarding early customers now. Join the waitlist to get priority access, influence the roadmap, and lock in founder pricing.

You're on the list — we'll be in touch soon.

We only use your email to contact you about Kuntur. No newsletters, no sharing.