Set limits on how your agents use money and track every action as it happens. No blind spots, no surprises. Just full visibility and control from the start.
Canopy gives your agents a wallet, governed by limits you set, attached to the agent code you already have. Three steps, three minutes.
Bring your own agent and set it up with a policy-bound wallet in minutes. Your agent gets a smart wallet on Base with gas sponsored.
Daily caps, human-in-the-loop approvals, allow-listed x402 service providers and more. Enforced at signing. Your agent can't transact outside the rule.
Drop the SDK into your agent's payment path. Every canopy.pay() call routes through your policy with live monitoring and a full audit trail.
import { Canopy } from "@canopy-ai/sdk"; const canopy = new Canopy({ apiKey: process.env.CANOPY_API_KEY!, agentId: process.env.CANOPY_AGENT_ID!, }); const result = await canopy.pay({ to: "agentic.market/anthropic", amountUsd: 0.10, });
If you're integrating, the docs are probably what you want. Here's the rest.
Canopy is the financial control plane for AI agents. We give your agents a stablecoin wallet they can spend from, a policy engine that decides what they're allowed to do with it, and an activity log that shows every payment they attempt, whether approved or blocked. The goal: let agents move money on your behalf without you losing sleep over it.
Anyone building or operating AI agents that need to spend money: trading bots, research swarms paying for data and compute, ops agents handling vendor payments and refunds, and commerce agents purchasing on behalf of users. If your agent has to send a payment, Canopy is the layer that makes that safe.
Canopy provisions a Privy-managed smart wallet as your organization's treasury. Agents are policy-enforced signers that spend from the treasury, so they don't hold funds themselves. Every signing request is evaluated against your policy server-side before Privy is asked to sign.
USDC on Base today. Other chains and stablecoins are not yet supported.
x402 today, the open HTTP-native payment standard for agents. Support for additional agentic payment standards is coming shortly.
Every signing request is evaluated server-side in a single atomic transaction against your policy. Today's rules cover daily spend caps, recipient allowlists, and approval thresholds that route over-limit transactions to a human reviewer. Violations are blocked at the API layer and recorded with a denial reason.
Transactions that violate policy are refused before they reach the chain, with the denial reason logged to the activity feed. You can pause an agent from the dashboard at any time to halt all further signing.
Pricing is in development. Reach out to the team for design partner terms.