Conto

How Conto works

When an agent goes to pay, Conto checks the request against your rules first, pauses anything that needs a second look, and keeps a record either way. Conto decides what clears, not what your agents buy. You choose whether Conto holds the wallet or the agent keeps its own key.

One control plane, from request to record

Every payment takes the same path. It comes in through your SDK or API, clears the same checks before it settles, and ends up in one record you can reconcile. The checks run before the money moves, not after.

Agents and operators

procurement
support
devops
operators

Integration surfaces

SDK
REST API
MCP
CLI
Conto Pay

Control plane

Conto

Every request is checked and scored before any funds move, then logged with full context.

Policy engine

Limits, trust, categories, and velocity evaluated before settlement.

Network intelligence

Trust and risk scores on every counterparty.

Approvals

Exceptions route to a human with full context.

Agent wallets

A governed wallet identity for each agent.

Settlement

Stablecoins
x402 pay-per-call

From wallets Conto manages or wallets controlled by the agent.

System of record

AgentPolicy resultApprovalCounterpartySettlementReceipt

Every payment reconciled with full context for finance, operations, and security.

One policy check, two execution paths

You pick who holds the keys. With a managed wallet, Conto sends the payment once it clears. With self-custody, the agent keeps its own key and sends the payment itself, while Conto signs off first and records the result after. The policy check and the audit trail are the same either way.

Managed wallet

Conto decides, the wallet sends · Conto holds the rules and the record

01 Request

Agent requests a payment

POST /payments/request
02 Policy check

Conto evaluates policy

approvereviewblock
03 If exception

Human approves

Routine spend skips this step.

04 Release

Wallet releases the funds

Once the request clears, the managed wallet signs and sends. Conto holds the decision and the record.

POST /payments/:id/execute
05 Record

Logged with full context

tx 0x9f2c…

policy · approval · settled

Agent-controlled wallet

Agent keeps its own keys · Conto authorizes, logs on confirm

01 Request

Agent asks to authorize

POST /payments/approve
02 Policy check

Conto evaluates policy

approvereviewblock
03 If exception

Human approves

Routine spend skips this step.

04 Execute

Agent signs with its own key

Conto returns a 10-minute approval. The agent sends the transaction.

05 Confirm + record

Agent reports the transaction ID

POST /payments/:id/confirm

tx 0x9f2c…

policy · approval · settled

Same payment context, same policy evaluation, same audit trail. Who signs and sends is where the path diverges.

Wallet sendsAgent signsException path

A control plane, not a custodian

Conto decides whether a payment goes through. The money itself moves on the rail, out of a wallet you manage or one the agent controls. Conto is never the one buying.

Conto is

  • A policy check before money moves
  • Trust and risk scores on every counterparty
  • Approvals with separation of duties
  • One record for finance and security

Conto is not

  • Your checkout or the buyer
  • A custodian, by default
  • A payment rail
  • A black box

Make your first policy-checked payment

Spin up an agent, wallet, and policy in the sandbox. It takes about ten minutes.