RapidfolioRapidfolio
Integrations

Integrations

Connect Rapidfolio to the services your financial operations already run on.

How integrations work

Integrations are the external services your procedures can call. When an AI agent executes a procedure, it can call Stripe to charge a customer, Plaid to fetch a bank balance, Onfido to run a KYC check, or Slack to send a notification — all as steps within the same workflow.

To use an integration, you create a connection first — an authenticated link to your account at that service. Once connected, any procedure can use it. You don't set up the authentication again for each procedure.

How to create and manage connections


Available integrations

Payments & Banking Stripe, GoCardless, Wise, Plaid, Alpaca

Compliance & KYC Alloy, Onfido, Flagright, ComplyAdvantage, Equifax, Experian, TransUnion, AIPrise, Middesk, Clearbit

Accounting Xero

Messaging & Project Management Slack, Linear, Jira

Custom & Private Custom API — connect any REST API using an OpenAPI spec Private Connection — run code inside your own infrastructure


Connecting a service you don't see here

If the built-in integrations don't cover your use case:

  • Custom API — paste in an OpenAPI spec and Rapidfolio auto-generates actions for every endpoint. No code required.
  • Private Connection — run your own code inside your network using the Connection SDK. The right choice for internal databases, core banking systems, or anything behind a firewall.

Environments

Connections are environment-scoped. Your sandbox Stripe account and your live Stripe account are separate connections. A procedure running in sandbox will never accidentally call your live Stripe account — it can only reach sandbox connections.

This means you can test the complete integration flow in sandbox, with real API calls and real responses, without touching any real customer data.


Human review for sensitive actions

Some integration actions automatically require human approval before executing — wire transfers, bulk operations, or anything your team has configured as high-risk. When the AI agent reaches one of these actions, execution pauses and the action appears in the Pending Reviews queue for a team member to approve.

Human Review guide

On this page