What is GFlow

GFlow Workflow Platform — approvals like the wind, flow at full speed. It orchestrates approval and automation in the same process, delivered in two forms: GFlow Engine, an embeddable Go approval engine (Apache-2.0 open source), and GFlow Platform, an out-of-the-box approval platform (commercial license, source code delivered).

Once approved, it runs itself. This is the fundamental difference from other approval engines: automation (invokes rule chains) and aiAgent (invokes AI agents) are engine-native nodes, orchestrated in the same DSL as approval nodes — AI pre-screens, humans approve, execution follows automatically. Countersign, add-sign, return, and other Chinese-style approval semantics form a complete foundation; see Chinese-style approval semantics.

Two Forms, One Core

GFlow Engine and GFlow Platform share the same DSL and data model; the only difference is the delivery form: embed the engine into your existing system, or deploy a complete, ready-to-use platform as is.

	GFlow Engine	GFlow Platform (GFlow Workflow Platform)
Form	Go library (embeddable)	Complete application (server + frontend)
Audience	Developers	Enterprises / business administrators
License	Apache-2.0 open source	Commercial license ¥7,500 · source code delivered
Repository	Gitee · GitHub	Delivered with the license

GFlow Engine is a lightweight approval workflow engine built on the RuleGo rule engine. Process definitions reuse the RuleGo rule chain DSL (JSON), and state such as approval tasks, process instances, and history archives is persisted by the engine to a relational database — no separate process middleware to deploy.

The DSL is a JSON-format, BPMN-like approval flow; it does not parse BPMN 2.0 XML.

GFlow Platform adds everything an enterprise needs on top of the engine to put it to real use. At its core are four self-developed components: the Process Designer (tree-based), the Form Designer (gform-designer), the Rule Chain Designer (visually orchestrating automation), and the AI Agent (AI approval / skills / agent management). Plus a frontend application for initiate / todo / done / CC / statistics, an admin backend for organization structure and multi-tenancy, automation orchestration, monitoring, and notifications. The frontend, the four designers, the AI agent, and automation are all built in-house — no third-party black boxes.

Core Features

• Rule chain as process: The process DSL reuses RuleGo rule chains, and native nodes such as the switch conditional gateway and fork/inclusive/join parallel branches can be orchestrated directly. A process is a rule chain; a rule chain is a process.
• Complete Chinese-style approval semantics: OR-sign (or), countersign (parallel/sequential, all-votes/majority/ratio/vote-count), dynamic add-sign/remove-sign, transfer, delegate, claim/grab, return to the previous node, withdraw, suspend/resume, and overdue reminders.
• 7 core tables: Runtime and history are kept on separate tracks; the data model is clear at a glance.
• Multi-tenant: tenant_id isolation across the entire chain, with rule chain execution pools partitioned by tenant.
• Pluggable identity system: Implement IdentityService to connect your real users/roles/departments (approvers resolved by role, department, or multi-level manager).
• Pluggable database dialects: PostgreSQL / MySQL are built in; other databases can be adapted on your own through the DialectProvider dialect extension point.
• Process definition versioning: The same process_key keeps multiple published versions with an increasing version; existing instances keep running on the old version.
• Lightweight deployment: No mandatory external middleware (Redis distributed locking is optional), a single Go binary — well suited for embedding into existing applications.

Who It Is For

• Teams adding approval capabilities to an existing system: import one Go library, create 7 tables, and keep your business code as is.
• Enterprises that want a complete approval platform: deploy the GFlow Workflow Platform directly — a business administrator can configure a first process in 30 minutes.
• Xinchuang (IT application innovation) / domestic-technology scenarios: a single Go binary with no JVM dependency — independent and fully under your control.

Live Demo

The demo environment is a complete, finished GFlow Platform deployment (with test data):

URL: http://8.134.32.225:8081 · Account: admin / admin123

Demo users: wangqiang / demo123456, zhangwei / demo123456 (they go through the multi-level approval paths).