Introduction

Orchesty 3.0 is a developer-first integration and orchestration platform. You write small pieces of code (workers), wire them into topologies, and Orchesty takes care of running them at scale, persisting every message, retrying failures, exposing metrics, and giving you a UI to operate the whole thing.

The 3.0 line keeps the same SDK contract that you may know from 2.x. The platform around it (Admin UI, deployment story, AI bootstrap) is significantly bigger.

When Orchesty is for you #

Orchesty fits projects that have at least one of:

• Two or more systems that need to stay in sync (e-shop and ERP, CRM and warehouse, IdP and downstream apps).
• Multi-step processes where you can't lose a message if a downstream API is temporarily down.
• Custom business logic that doesn't fit a no-code iPaaS but isn't worth a bespoke microservice either.
• Compliance or sovereignty constraints that need on-prem or private-cloud control.

If you only need a one-shot HTTP call, Orchesty is overkill. If you need durable, observable, retry-safe message flow with custom code, it is built for you.

What you build with Orchesty #

You build topologies, which are graphs of nodes connected by queues. Each node is a small piece of code (a worker handler) that:

• Receives one message,
• Does something with it: HTTP call, transformation, decision,
• Returns a result that the platform routes to the next node.

The Admin UI lets you draw, deploy, monitor, and debug topologies. The SDK lets you implement the nodes themselves in Node.js or PHP.

What's new in 3.0 #

• AI bootstrap for the worker layer. A starter project ships with rules and prompts that let an AI agent scaffold real worker code from a one-line goal. See Using AI to bootstrap.
• Two on-ramps: a slim worker setup that talks to an existing instance, and the full local stack with the Admin UI. See Choose your setup.
• Cleaner Admin UI focused on operability: dashboards, heatmaps, the Trash inbox for failed messages.
• Reorganized docs: Concepts (the model), Development (how to build), Reference (per-SDK API).

Where to go next #

1. Decide between the worker-only and full-stack setup in Choose your setup.
2. Read Architecture for the platform model.
3. When you're ready to ship, browse Development and consult Reference for exact signatures.