Shadow AI Discovery · AI Monitoring · Secrets Detection

AI Detection & Response for your entire workforce

Detect and secure the AI apps your workforce and developers actually use. Identify and stop Shadow AI. Monitor what happens inside each AI session. Catch secrets and sensitive data in prompts and redact them without stopping work. Track spend, enforce compliance, and get work done.

Book a Demo Start Free Trial Watch the 60-second tour

See It In Action

Kraitos AIDR in 60 Seconds

See every AI tool in your fleet, find the secrets in AI sessions, and prove compliance, all in one minute.

The Problem

AI Tools Are Your Biggest Blind Spot

AI coding agents execute shell commands, access files, install packages, and write production code - with virtually no oversight.

No inventory. Developers install Claude Code, Cursor, Copilot, Windsurf, and a dozen others on their own. IT doesn't know what's running.
No data controls. API keys, credentials, and proprietary code leak into AI prompts with no detection or prevention.
No cost visibility. Finance asks "what are we paying for AI APIs?" and nobody can answer.
No audit trail. When an AI agent deletes a production file or installs a vulnerable package, there's no record.
No compliance evidence. SOC 2, ISO 27001, and NIST auditors are asking about AI governance - and most teams have no answer.

Traditional EDR platforms detect malware and lateral movement. They have zero visibility into what AI agents are writing, what packages they install, what secrets they encounter, or whether their activity complies with your policies.

The Platform

One Agent. Four Pillars.

Kraitos AIDR deploys as a single lightweight agent - under 50 MB memory, less than 1% CPU - covering AI governance, data protection, threat detection, and asset management in one install.

Monitor

Real-time visibility into every AI tool, every session, every token, every dollar across your fleet.

AI Tool Discovery
Session Intelligence
Cost & Usage Analytics
Live Floor
Data Egress
Local AI Detection

Explore Monitor

Protect

Enforce policies, detect secrets, scan for threats, and produce audit-ready compliance evidence.

Policy Engine
Secret & DLP Detection
Threat Detection
Compliance Automation
AI Profiles

Explore Protect

Respond

Automated and human-gated response actions with SOAR-lite playbooks for AI-related incidents.

Isolate / Kill / Expire
Playbook Builder
Quarantine Manager
Approval Workflows

Explore Respond

Assets

Complete endpoint inventory: hardware, software, browser extensions, and AI tool configurations.

Device Inventory
Software & Hardware
Browser Extensions
Repository Discovery
Dependency Audit

Explore Assets

Kraitos AIDR executive overview showing fleet KPIs, endpoint protection status, token usage over time, and AI tool distribution — Executive overview - active devices, AI tools detected, tokens and cost, open alerts and secrets, engine health, and tool distribution across the fleet.

AI Tool Discovery

Find Every AI Tool on Every Endpoint - Known or Unknown

Shadow AI is the number one risk CISOs cite when it comes to AI adoption. Developers install tools without IT approval, connect to unauthorized LLM APIs, and run local inference servers that never touch the corporate network.

Kraitos AIDR's five-layer detection architecture closes this blind spot. It identifies 48 AI tools by name, monitors network connections to 115 AI service endpoints across 69 providers, and uses behavioral heuristics to catch tools that don't have signatures yet - including custom scripts that call LLM APIs directly.

Discovery	Filesystem scanning for config directories, VS Code and JetBrains extensions, and process names - 48 AI tools detected by name
Network	Connection monitoring identifies 115 AI service endpoints across 69 providers, plus 7 local inference servers, by destination
Behavioral	Behavioral heuristics recognize how AI tools act - catching unknown and custom tools no signature list has heard of
Telemetry	Deep session telemetry for Claude Code - full conversation replay, token tracking, MCP visibility
Dependency	Dependency monitoring across 9 package formats, with known vulnerabilities flagged automatically

Per-device AI tool inventory listing each detected tool with version, first seen, and last seen dates — Per-device AI tool inventory - every tool found by name or by network destination, with version and first and last seen.

Supported Tools: Claude Code, Codex (OpenAI CLI), Aider, OpenCode, OpenHands, Devin, Factory Droid, Amp (Sourcegraph), Goose (Block), Crush (Charm), Qwen Code, Kilo Code, Roo Code, Cline, OpenClaw, Kiro, Google Antigravity, Warp, Cursor, Windsurf, Codeium, GitHub Copilot, Cody (Sourcegraph), Continue, Tabby, Tabnine, Supermaven, Amazon Q Developer, Gemini Code Assist, Zed, Augment Code, Blackbox AI, Pieces for Developers, Trae, Qodo Gen, JetBrains AI Assistant, JetBrains Junie, Claude Desktop, ChatGPT Desktop, Microsoft Copilot, Grok (xAI), Ollama, LM Studio, Jan.ai, GPT4All, Msty, AnythingLLM, Open WebUI
Local Inference: Ollama, LM Studio, Jan.ai, GPT4All, llama.cpp, text-generation-webui, vLLM
Cloud APIs: 115 AI service endpoints across 69 providers, including OpenAI, Anthropic, Google, AWS Bedrock, Azure OpenAI, Groq, Mistral, Cohere, and DeepSeek
WSL Support: Full cross-boundary detection for Windows Subsystem for Linux environments

Deep Session Intelligence

Full Conversation Replay - Every Prompt, Tool Call, and File Access

Kraitos AIDR reconstructs complete Claude Code sessions with every turn: user prompts, assistant responses, tool invocations, file reads and writes, shell commands executed, and MCP server connections - all with per-turn token counts and cost.

When an incident occurs - a leaked credential, a deleted production file, a malicious package install - you have the full forensic record. Not just “something happened” but exactly what the AI did, what the developer asked, and what changed.

Complete conversation timeline - user prompts, assistant responses, and every tool call in sequence
File operations tracked - every read, write, and edit performed by the agent, with path and context
Shell command logging - full command text, working directory, execution context
MCP server connections - which external tool servers the AI reached and what it sent
Subagent relationships - parent/child session mapping when the AI spawns background agents
Cost attribution - per-turn token counts (input, output, cache read, cache write) with model-aware pricing

AI session list grouped by tool, device, and day, each row showing risk score, session count, secret findings, and cost — AI sessions grouped by tool, device, and day - each group leads with risk score, session count, secret findings, and cost.

Secret & Data Protection

Real-Time DLP for AI Conversations - Detect, Redact, Allow

AI coding agents routinely encounter credentials in codebases - API keys in config files, database passwords in environment variables, private keys in deployment scripts. Without protection, these secrets end up in prompts sent to cloud LLM providers, creating credential exposure events that can trigger breach notification requirements.

Kraitos AIDR scans AI conversations in real time with 37+ detection patterns across five data categories. When a secret is detected, the platform can redact the sensitive content directly from the session file while allowing the conversation to continue - protecting the credential without disrupting the developer's workflow.

SecretsPCIPHIPIIFinancial

37+ detection patterns across cloud providers, source control, AI providers, and payments
Infrastructure credentials - private keys, JWTs, database connection strings, Redis URIs
PII and regulated data - SSNs, emails, phone numbers, medical record patterns, card numbers
Unknown-format detection - entropy analysis catches high-randomness secrets that match no known pattern
Context-aware attribution - was it a user prompt, an assistant response, or a tool call output?
Live credential verification - probe the provider to learn whether the leak is live, revoked, or unknown
Tracked rotation lifecycle - mint a replacement, revoke the leak, and re-verify that it is dead

Data loss prevention dashboard with an exposure matrix of data class against severity and framework exposure — DLP command center - exposure matrix of data class against urgency, with live-credential marking and framework exposure.

Secret finding detail panel showing exposure duration, redaction diff, data class, source, and remediation actions — Finding detail - how long the value has been exposed, the redaction diff, where it appeared, and the verify, redact, and resolve actions.

Data Egress Monitoring

Know Where Your Data Goes - Sanctioned AI, Shadow AI, and Everything Between

Every AI tool on every endpoint is making outbound connections - to cloud LLM APIs, to MCP servers, to cloud storage, to destinations your security team has never evaluated. Kraitos AIDR classifies every outbound AI-related connection and gives you the controls to set data boundary policies and get alerted on them.

The egress console categorizes every destination as sanctioned, shadow, cloud storage, unknown, or not allowed - and lets you reclassify it. Promote a destination to sanctioned, mark it as not allowed, or route the decision through the policy engine for automated alerting.

Sanctioned AIApproved provider

Shadow AIUnapproved tool

Cloud StorageFile destination

UnknownAwaiting review

Not allowedMarked by an admin

AI egress console listing destinations with provider, sanctioned or shadow classification, connection volume, and sanction and not-allowed labels — AI egress - every destination classified sanctioned or shadow, ranked by connection volume, with sanction and not-allowed labels.

Capabilities

Everything Else in the Same Agent

No second install, no bolt-on console, no per-module pricing. Every capability below ships in the agent you already deployed.

AI Profiles

Group Policy for AI. Define CLAUDE.md, .cursorrules, Copilot instructions, and MCP allowlists centrally, then push them to every managed repo on every machine.

Template library
Discover & assign
Drift detection
Hierarchical merge

Learn more

Policy Engine

Watch everywhere, or block in Claude Code - rules cover tools, models, data, spend, and dependencies. Policies stream to agents in under a second and evaluate locally in sub-millisecond time.

Monitor & enforce modes
4 scope levels
23+ templates
Audited exceptions

Learn more

Compliance Automation

53 controls across 7 frameworks, with evidence generated continuously as developers work - not assembled the week before an audit.

Signed evidence bundles
Drift detection
Control attestation
Auditor export

Learn more

Endpoint Protection

Seven detection engines ship in the same agent: AI discovery, secret scanning, native malware, YARA, IOC matching, Sigma behavioral, and file integrity monitoring.

6.5M+ malware signatures
MITRE ATT&CK mapping
Posture checks
Compliance scoring

Learn more

Response & Playbooks

Isolate a device, kill a process, or expire a leaked credential - automatically or behind a human approval gate, with a full audit trail on every action.

Device containment
SOAR-lite playbooks
Approval workflows
Encrypted quarantine

Learn more

Cost & Usage Analytics

Per-developer, per-team, per-model token and cost tracking with 30-day trends, budget alerts, and chargeback-ready exports for finance.

Model-aware pricing
Budget thresholds
Adoption metrics
Finance exports

Learn more

Detection Engines

Seven Engines. One Install.

EDR-class protection ships alongside AI governance. Every engine keeps itself current - signatures update automatically, with no restart and no maintenance window - and feeds the unified device compliance score.

Detection engines
Engine	Function	Method
AI Tool Discovery	Find every AI coding tool on every endpoint	Filesystem, network, behavioral, and deep telemetry layers
Secret Scanner	Detect credentials and sensitive data in AI conversations	37+ patterns plus entropy analysis for secrets no pattern knows about
Native Malware Scanner	Detect malicious binaries and files	6.5M+ malware signatures, refreshed automatically
YARA Scanner	Match files against YARA rules	A yara-x engine with a compiled rule bundle, built into the agent with nothing extra to install
IOC Matcher	Match against curated threat intelligence	File, domain, and IP indicators from continuously updated threat feeds
Sigma Behavioral	Detect suspicious endpoint behavior	Behavioral rules with full MITRE ATT&CK mapping
File Integrity Monitor	Detect modification of critical system files	Tamper baselines with real-time change alerts

SOC 2
ISO 27001
NIST 800-53
PCI DSS v4.0.1
HIPAA
NIST AI RMF
EU AI Act

53 controls mapped across seven frameworks, with evidence generated continuously as developers work.

Deployment

Fast Deployment. Zero Developer Friction.

No infrastructure to manage. No reboot. No developer interaction. The agent runs silently - no popups, no configuration screens, no prompts.

Step 01
Create an enrollment code
Generate a short enrollment code in the dashboard. Codes carry their own constraints: OS restrictions, CIDR allowlists, team auto-assignment, single- or multi-use, and an expiry window.
Step 02
Install the agent
Download the signed Windows MSI or the macOS and Linux installer from the dashboard and run it with the code, or push it through your existing endpoint management tooling. No reboot, no imaging, no developer interaction.
Step 03
Devices report instantly
AI tools, sessions, posture, and inventory flow into the dashboard within seconds of the agent starting. A short guided tour on first sign-in walks you through deployment, the Live Floor, AI Sessions, Inventory, and Compliance.

Memory: < 50 MB
CPU: < 1%
Platforms: macOS · Linux · Windows
Updates: Automatic

Buy direct, or subscribe through AWS Marketplace and pay on your AWS bill.

Why Kraitos

Three Markets Each Solve a Third of the Problem

EDR sees malware but not AI. AI security platforms see the model layer but ship no endpoint agent. Compliance automation sees neither. Kraitos AIDR is all three in one agent, on one per-endpoint line item.

AI Governance + Endpoint Security

Full AI session visibility, data loss prevention, and EDR-class endpoint protection in a single agent. No bolt-ons, no integrations, no second install.

One agent
Seven detection engines
Under 50 MB
No second console

Local-First Policy Enforcement

Policies evaluate on the endpoint in sub-millisecond time - no cloud round-trip, no latency, no single point of failure. Developers do not experience a slowdown.

Sub-ms evaluation
Policy push under 1s
Works offline
Monitor & enforce modes

Full Conversation Intelligence

Not just prompts - every tool call, file access, shell command, MCP connection, and subagent relationship, with per-turn token counts and cost.

Session replay
Shell command log
MCP destinations
Cost attribution

AI Profiles - Group Policy for AI

Centrally manage CLAUDE.md, .cursorrules, MCP allowlists, and memory packs across every repo on every machine, with drift detection and reconciliation.