One model for entity state, workflows, events, and transactions.

Cyoda is an open-source entity database management system (EDBMS) and workflow runtime. You define your entity types — the fields they carry, the states they can be in, and the valid transitions between states. Cyoda enforces those rules at the API boundary, records every state change immutably, and lets you query the history of any entity at any point in time. It runs as a single binary with no distributed cluster required.

An entity database management system (EDBMS) is a runtime that manages the full lifecycle of typed, stateful business objects — entities. Where a standard database stores rows, an EDBMS tracks state transitions, enforces lifecycle rules, and maintains a temporal audit trail. Cyoda adds a workflow engine to this model: transition rules are explicit, invalid transitions are rejected at the API boundary, and the complete state history is queryable.

Most backend systems need to track the state of business objects over time — orders, users, requests, approvals, jobs. Without a structured model, state logic ends up scattered across application code and database columns with no enforcement and no history. Cyoda gives each entity a defined lifecycle with enforced rules. Your application code drives transitions; Cyoda ensures they are valid and records them.

Systems where entities change state over time and where the history of those changes matters: workflow automation, order management, approval pipelines, event-driven backends, audit-sensitive applications, and anywhere you would otherwise build a hand-rolled state machine with an audit log.

Most backend systems require separate tools for each: a workflow engine, a state machine library, an event bus, and a transaction manager, connected by integration code. Each connection point is a source of inconsistency, hidden state, and operational complexity. Cyoda runs these capabilities as one runtime. Entity state, lifecycle transitions, event-driven processing, and transactional consistency all operate within the same model — with no integration layer between them.

AI tools generate better code when the system model is explicit and structured. Because Cyoda defines entity types, valid states, and transition rules as queryable schema rather than implicit application logic, AI tools can reason about system behavior from the model rather than inferring it from scattered code. The result is more accurate code generation, more precise debugging assistance, and a lower risk of incorrect state handling in generated output.

SQLite-backed local storage is the default for a fresh local install. In-memory mode stores entity state in process memory and resets when the process stops, which is useful for fast functional tests. PostgreSQL stores entity state durably for service deployments that need an external database.

Run it yourself: install the open-source binary and run Cyoda on your own machine or infrastructure. Free, Apache 2.0, no account required. Cyoda Cloud: the hosted SaaS option at ai.cyoda.net — managed infrastructure, no installation required. Enterprise Cyoda: larger-scale deployment with enterprise support, SLA, and dedicated engagement for organisations with advanced operational requirements. All three options share the same core API and entity model.

Java, Python, Go, and Kotlin via gRPC client libraries. Any language with a gRPC implementation can connect to the Cyoda API directly.

The source code is at github.com/Cyoda-platform/cyoda-go. Issues, releases, and contributions are managed there.

Full documentation is at docs.cyoda.net. The install-and-first-entity guide covers installation, local startup, importing an entity workflow, creating an entity, invoking a lifecycle transition, and reading state back.