Skip to content
Documentation

Build on the Agent Banking Interface.

The ABI is designed so authorised agents can hold, move, and settle value across fiat and stablecoin rails within governed mandates. Everything here is the entry point for integration.

Request documentation accessDeveloper overview
Core Concepts

The building blocks of agent-native banking

These are the foundational concepts that shape how the ABI works. Understanding them is the fastest path to a working integration.

api

Agent Banking Interface (ABI)

The programmatic surface through which authorised agents and software operators interact with the Vayan operating layer. Designed for tool-calling runtimes, treasury platforms, and governed settlement workflows.

verified_user

Mandates

Every ABI operator registers a mandate that defines transaction boundaries: maximum amounts, permitted rails, counterparty types, and human escalation thresholds. Execution outside the mandate is structurally blocked.

swap_horiz

Dual-Rail Settlement

Rail A handles fiat. Rail B handles permitted stablecoins. The ABI routes across both based on policy, capability, and the operator's mandate configuration. One instruction, two possible paths.

fact_check

Compliance Metadata Tag

Every regulated transaction generates a structured compliance record: screening results, mandate references, policy decisions, and ledger hashes. Evidence is a byproduct of execution, not a post-hoc reconstruction.

hub

Multi-Agent Consensus (MAC)

No single agent executes alone. Independent control layers — compliance, treasury, policy — must reach consensus before regulated settlement proceeds. The model is structural enforcement, not procedural sign-off.

visibility

Observer Agent

A read-only supervisory interface for authorised oversight bodies. Exposes live ledger views, compliance records, reserve visibility, and system-health indicators without write access.

Integration Patterns

One tool call. Any framework.

The ABI is designed to fit into the tool-calling pattern of any major agent framework. Below are illustrative integration shapes.

LangChain / LangGraphStructuredTool
from langchain.tools import StructuredTool
from vayan_sdk import VayanClient

client = VayanClient(api_key=VAYAN_API_KEY)
settle = StructuredTool.from_function(
    name="submit_settlement",
    func=client.submit_instruction,
    args_schema=SettlementRequest
)
OpenAI Function CallingJSON Schema Tool Spec
tools = [{
  "type": "function",
  "function": {
    "name": "submit_settlement",
    "parameters": {
      "type": "object",
      "properties": {
        "amount": { "type": "number" },
        "currency": { "type": "string" },
        "rail": { "enum": ["fiat","stablecoin","auto"] }
      }
    }
  }
}]
Anthropic Tool UseTool Definition
tools = [{
  "name": "submit_settlement",
  "description": "Submit a governed settlement instruction.",
  "input_schema": {
    "type": "object",
    "properties": {
      "amount": { "type": "number" },
      "currency": { "type": "string" },
      "rail": { "type": "string" }
    },
    "required": ["amount", "currency"]
  }
}]
REST / Direct HTTPOpenAPI 3.1
curl -X POST https://api.vayan.finance/v1/settlements \
  -H "Authorization: Bearer $VAYAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "amount": 50000,
    "currency": "USD",
    "rail": "auto",
    "beneficiary": "acct_recipient_id"
  }'

Illustrative examples only. SDK and final schemas are still in development.

Getting Started

From request to running integration

1
Request access

Write to dev@vayan.finance with your use case, framework, and team context. We respond with sandbox credentials and onboarding materials.

2
Configure your mandate

Define transaction limits, permitted rails, counterparty types, and escalation thresholds. The sandbox enforces mandate boundaries identically to production.

3
Integrate the ABI

Use the SDK, REST API, or your preferred agent framework. The sandbox supports the same endpoints, schemas, and compliance flows as the production environment.

4
Test settlement flows

Submit instructions, observe compliance validation, inspect metadata tags, and test mandate boundary enforcement. Mock counterparties and test accounts are provided.

5
Review with the team

Walk through your integration, compliance records, and edge cases with the Vayan developer team before requesting production credentials.

Documentation Library

Reference materials

Documentation is shared with qualified partners and early-access developers. Request access to receive the materials relevant to your integration.

menu_bookAvailable by request

ABI Reference

Endpoint shapes, authentication approach, settlement flows, mandate registration, and compliance metadata structures.

scienceAvailable by request

Sandbox Setup Guide

Provisioning, environment variables, mock accounts, test counterparties, and integration constraints before authorised launch.

integration_instructionsAvailable by request

Agent Integration Guide

Framework-specific examples for LangChain, OpenAI, Anthropic, CrewAI, AutoGen, REST, and MCP-style integration.

descriptionAvailable by request

Compliance Record Specification

Schema definitions for the Compliance Metadata Tag, event ledger records, and structured evidence outputs.

tuneAvailable by request

Mandate Configuration Guide

How to define operator mandates, escalation thresholds, rail preferences, and counterparty restrictions.

shieldComing soon

Security Model

Authentication, key management, rate limiting, and the trust model between operator, agent, and the Vayan operating layer.

Need a direct walkthrough?

If you are integrating the ABI into a treasury system, payment workflow, or agent runtime, we can walk through the mandate model, compliance lifecycle, and current sandbox-planning assumptions with your team.

Contact developer supportRead the developer overview