Skip to main content

The Problem

AI agents are increasingly autonomous — they browse the web, call APIs, and make decisions. But when an agent needs to pay for something, the options are limited and dangerous:
  • Giving agents private keys means a single prompt injection or bug can drain the wallet entirely.
  • Human-in-the-loop for every payment defeats the purpose of autonomous agents.
  • Traditional wallet infrastructure was designed for humans clicking buttons, not machines making thousands of decisions per hour.
Storing private keys in agent memory or environment variables is the number one cause of AI agent fund loss. A compromised agent with a private key has unlimited access to all funds.

What AgentWallex Does Differently

No Private Key Exposure

AgentWallex uses 2-of-3 threshold MPC signing via Paratro. The full private key is never reconstructed — not in memory, not on disk, not anywhere. Even if your agent or API key is compromised, the attacker cannot sign transactions alone.

Policy Engine as a Safety Net

Every transaction passes through a configurable policy engine before signing. Set per-agent limits on:
  • Transaction amount — cap individual payments
  • Daily and monthly totals — rolling spending limits
  • Allowed addresses — whitelist trusted recipients
  • Token restrictions — limit to stablecoins only
  • Velocity controls — prevent runaway transaction loops
  • Time-based rules — restrict to business hours
  • Human approval — route high-value transactions to human review

Infrastructure Safety Net

Even if your business-layer policies are misconfigured, Paratro enforces hard limits at the MPC signing level that cannot be overridden by API calls:
ControlDefault
Absolute daily cap$50,000
Anomaly detectionML-based, always on
Emergency freezeInstant, always available
Cool-down period10 min after freeze

x402 Native Support

AgentWallex natively supports the x402 machine-to-machine payment protocol. Your agents can automatically negotiate and pay for API access using HTTP 402 headers — no human intervention needed.

Who Uses AgentWallex

AI Agent Builders

Teams building autonomous agents that need to transact on-chain — paying for data, APIs, or services.

API Providers

Services that want to monetize their APIs using x402 micropayments with instant on-chain settlement.

Crypto-Native Teams

Projects that need secure, policy-controlled wallet infrastructure for programmatic on-chain operations.

Key Differentiators

FeatureTraditional WalletsAgentWallex
Key storageSingle private key2-of-3 MPC shards
Access controlAll-or-nothingPer-agent policies
Payment protocolsManual tx signingx402 automatic negotiation
Spending limitsNone (or contract-level)Configurable per agent
Human oversightNone or fullThreshold-based routing
Audit trailBlock explorer onlyFull API + policy logs
Multi-chainPer-walletSingle API, many chains

Next Steps

Quickstart

Get your first agent transacting in under 5 minutes.

Security Architecture

Deep dive into the four security layers.