Skip to content
0x1e.ai
Platform

A control plane for agentic financial execution.

0x1e.ai is built as a set of cohesive modules that turn agent intent into controlled execution — identity, policy, vault, approval, rail, ledger, audit, webhooks. Each one is independently addressable, all of them wired together.

Start integratingRequest early access

Modules

Eight modules. One contract.

Each module owns a clear responsibility. The boundaries are designed so the system stays auditable and the rails stay swappable.

Agent Registry

Identity, ownership, state machine, and rotation primitives for every registered agent.

KYA Service

Submission, profile, review, and decision workflow for Know Your Agent compliance.

Policy Engine

Per-agent caps, velocity limits, allow-lists, and merchant rules evaluated at pre-check.

Programmable Vaults

Multi-currency balances with available, reserved, and clearing partitions.

Approval Orchestrator

Owner notification, confirmation handling, timeout enforcement, and atomic reservation.

Payment Rail Adapter

Rail-agnostic execution interface. Stripe today, additional rails behind the same contract.

Ledger & Audit Trail

Append-only double-entry ledger and append-only audit log. Reconciliation built-in.

Webhooks

Signed delivery of proposal, payment, and refund events to downstream consumers.

Flow

Seven stages from intent to record.

Every proposal walks the same path. No skipped steps, no shortcuts.

  1. 01ProposalAgent submits
  2. 02ValidationPolicy + KYA
  3. 03ApprovalOwner confirms
  4. 04ReserveVault locks funds
  5. 05ExecuteRail pays
  6. 06ReconcileCommit or release
  7. 07AuditAppend-only record

Principles

Six rules the platform never breaks.

They're explicit because they're load-bearing. Every architectural choice points back to one of them.

01

Always confirm first

No silent automation. Every spend action requires an explicit owner confirmation before execution.

02

Reserve on approval

Approved funds are moved to reserved atomically. Capacity is honest from the moment of approval.

03

Multi-currency ready

Balances and proposals are denominated per currency. No implicit conversion at the platform layer.

04

Agent-specific caps

Spend capacity is computed against agent-level limits, not just owner balance. One bad agent does not drain the wallet.

05

Append-only audit

Ledger and audit logs are write-once. Corrections happen through compensating entries, not edits.

06

Rail-agnostic execution

The platform is not the rail. Payment providers sit behind an adapter so they can be swapped or stacked.

Example flow

An agent wants to buy $500 in API credits.

The timeline on the right is exactly what happens. Nothing is elided. Every entry is recorded append-only and reflected in ledger and audit.

  • agent · agt_4f2c · billing-assistant-v2
  • owner · acme.industries
  • total elapsed · 48.6s
  1. +0s
    Proposal createdcreated

    agent: agt_4f2c · amount: 500.00 USD · merchant: API Provider

  2. +0.1s
    Policy check passed

    agent cap 500/1000 USD · velocity 0/3 today · merchant allow-listed

  3. +0.2s
    Owner notifiedpending owner

    channel: telegram · 4h confirmation window opened

  4. +47s
    Owner approvedapproved

    via signed callback · idempotency_key honoured

  5. +47.0s
    Funds reserved

    owner_available −500.00 → owner_reserved +500.00 (entry_type: reserve)

  6. +48s
    Rail executed paymentexecuted

    stripe rail · charge_id: ch_…3aa1

  7. +48.4s
    Reserve committed

    owner_reserved −500.00 → system_clearing +500.00 (entry_type: commit)

  8. +48.5s
    Webhooks dispatched

    payment.succeeded · proposal.executed

  9. +48.6s
    Audit trail closed

    9 append-only entries · proposal lifecycle complete

Start integrating agents you can trust.

The full platform is documented and available behind early access. Get a key, sign your first proposal, ship.

Request API accessView developer docs