Features Pricing Use Cases Scoring FAQ About
Sign in Get Started
🛡️   Governance for AI Agents in Web3

Secure Your
Autonomous Future

"Witz" — Sacred Mountain, the boundary between worlds

Real-time monitoring, auditing, and containment for autonomous AI agents on Base Network. LSTM anomaly detection, Rule Engine, and on-chain audit trail.

⛓️  Base Network 🤖  AI-Powered 🔒  Web3 Native
<10s
Alert Latency
4
Detection Layers
0–100
Risk Score
24/7
On-chain Monitoring

Everything You Need to
Secure AI Agents

Comprehensive governance tools designed specifically for autonomous AI agents operating in Web3.

AI-Powered
🧠

Dynamic Risk Score

Real-time 0–100 score based on transaction history (40%), market context (25%), network behavior (20%), and macro factors (15%).

<10s Response

Circuit Breaker

Automatic containment of anomalous agents in under 10 seconds. Pause, investigate, and resume with full confidence and audit trail.

Live Data
📊

Real-Time Dashboard

Live monitoring of all your agents with instant alerts via Telegram, Discord, and Email. War Room mode for enterprise incidents.

On-Chain
🔗

Immutable Audit Trail

All decisions and alerts hashed and stored on Base Network via AuditRegistry smart contract. Tamper-proof, verifiable forever.

Rules + ML
🔬

Hybrid Intelligence

Deterministic Rule Engine for known threats combined with per-agent LSTM Autoencoder for novel anomalies. Best of both worlds.

Decentralized
🌐

Sentinel Network

Decentralized observation network (Phase 2) providing global coverage and redundant validation with aligned economic incentives.

$2.71B
Losses Prevented in 2025
82:1
Agents vs Humans Ratio
6%
Have Advanced AI Security
143
Tests Passing

Up and Running in 3 Steps

No infrastructure changes required. Connect your agent, watch the data flow.

1

Register Your Agent

Point WITZ Guardian at your agent's wallet address. We start scanning on-chain transactions immediately via Alchemy WebSocket.

POST /api/agents/register
{
  "wallet": "0x...",
  "name": "My Agent",
  "network": "base"
}
2

Monitor & Detect

Four detection layers analyze every transaction in real time: Rule Engine → LSTM Anomaly → KYA Protocol → On-chain Audit Writer.

GET /api/agents/{id}/risk-score

{
  "score": 72,
  "ml_mode": "shadow",
  "anomaly": true
}
3

Alert & Contain

Receive instant alerts on Telegram, Discord, or Email. Trigger the Circuit Breaker automatically or manually via API or Dashboard.

PATCH /api/agents/{id}
{
  "is_monitored": false,
  "reason": "anomaly_detected"
}

Simple, Transparent Pricing

Choose the plan that fits your needs. Upgrade or downgrade anytime.

Monthly
Annual Save 20%

Free

Perfect for individual developers getting started

$0 /month
  • 3 agents monitored
  • 100 alerts/month
  • Predefined rules
  • Telegram alerts
  • Community support
  • Custom rules
  • API access
  • Circuit Breaker
Get Started

Enterprise

For protocols and institutions requiring SLA

$199 /month
  • Unlimited agents
  • Unlimited alerts
  • Custom rules
  • API access (10K req/h)
  • Circuit Breaker
  • Email + all channels
  • SLA 99.9%
  • Dedicated support
Contact Sales

Need a custom solution? Contact us for tailored enterprise pricing.

🌐   Coming Soon

The Sentinel Network

Decentralized observation that scales coverage without scaling costs. Intelligence stays centralized and protected — observation goes global.

Traditional Centralized

🏢
Single server observes all
📈
Scales with infrastructure cost
📍
Geographic limitations

Witz Hybrid Model

🛡️
Centralized intelligence + decentralized observation
Scales with network participation
🌐
Global redundant coverage
🌍

Global Coverage

Distributed observers across regions provide redundant validation and eliminate single points of failure.

👥

Community Powered

Anyone can run a Sentinel client and earn rewards for contributing valuable observation data.

🪙

Fair Rewards

Sentinels earn proportional to the verified utility of their data, not just volume submitted.

🎯

Improved Accuracy

Multiple independent observations increase confidence and reduce false positives significantly.

Sentinel Reward Structure

~$0.001/tx
Base Observation
Verified transaction report
~$0.10–1.00
Anomaly Detected
Confirmed by AI model
5–10% of sub
Alert Triggered
Led to client notification
1–2% of value
Containment Success
Prevented loss (capped)

20–30% of subscription revenue goes to the Sentinel reward pool. Rewards distributed proportionally based on verified utility.

👁️ Get Notified When Sentinel Launches

Built on Best-in-Class Infrastructure

⛓️
Base Network
L2 Blockchain
Alchemy
RPC + WebSocket
🗄️
Supabase
Database + Auth
🔥
PyTorch
LSTM Autoencoder
🤖
Virtuals
Agent Protocol
🚀
FastAPI
Backend API

Start Protecting Your
AI Agents Today

Create your free account in seconds. No credit card required. Start monitoring your AI agents immediately.

Create Free Account → How Scoring Works

Free plan includes 3 agents, real-time alerts, and full dashboard access.

Stay Updated

Get the latest on AI agent security, product updates, and industry insights. No spam, unsubscribe anytime.

We respect your privacy. Unsubscribe at any time.

Frequently Asked Questions

What is WITZ Guardian?

WITZ Guardian is a governance and security platform for autonomous AI agents operating on blockchain. It provides real-time monitoring, anomaly detection using AI (LSTM + Rule Engine), and automated containment to protect your agents and assets.

Is WITZ a competitor to agent developers or frameworks?

No. WITZ Guardian is infrastructure that complements agent developers and frameworks like Virtuals, Olas, and Coinbase AgentKit. We don't build agents — we protect them. Think of us as the security layer that makes your agents trustworthy for users and institutions.

How does the scoring work?

Each agent receives a dynamic 0-100 risk score based on transaction patterns, behavioral anomalies, and market context. Our 6-layer detection stack (VAE-LSTM, Isolation Forest, CUSUM drift detector, graph analyzer, conformal thresholds, and adaptive calibration) analyzes every transaction in real-time.

How much does it cost?

We offer a free plan with up to 3 agents and real-time alerts. Pro plans start at $49/month for unlimited agents and advanced features. Enterprise plans include dedicated support and custom SLAs. See our pricing section above for details.

Is my data secure?

Yes. We only analyze publicly available on-chain data. Your private keys are never accessed. All detection decisions are recorded on an immutable on-chain audit trail via Base Network smart contracts, ensuring full transparency and accountability.

Which blockchains do you support?

Currently we support Base Network (Coinbase L2). Support for additional EVM-compatible chains is on our roadmap. Our architecture is chain-agnostic, making expansion straightforward.

What is the Sentinel Network?

The Sentinel Network (Phase 2) is a decentralized observation layer where community members run lightweight monitoring nodes. Sentinels earn rewards for validating agent behavior, creating a global security network with aligned economic incentives. Coming soon.