Platform
Under the hood: KotlinDynamicRuntime.
Gedra is built on KotlinDynamicRuntime, an open-source project whose central idea is simple: keep what an application is, its forms, rules, and workflows, as declarative configuration rather than as scattered code.
Forms are JSON Schema
Every form, field, and rule is defined in JSON Schema (draft 2020-12), with conditionals for questions that depend on earlier answers. The same definitions validate data in the browser, on the server, and in the APIs.
Workflows are configuration
Tasks, the forms each one edits, approvals, and locks are stored configuration. A client's configuration can be edited and tried in a sandbox, then published when it is ready.
Built for AI to configure
Configuration has a small, well-defined construction space, so AI can write it reliably, even smaller and cheaper models. The platform was designed with AI in mind from its first issue.
One language, server to browser
Kotlin runs on the server and, through Kotlin Multiplatform, in the browser, sharing one code base, so a rule is written once and enforced on both sides. We are also open to building for your frontend stack.
APIs fitted to each client
Endpoints are described by the same schemas as the data, so the APIs a deployment offers are documented, validated, and shaped to the systems it talks to.
Extensible, not forked
What is unique to a client is added as extensions and configuration, so a deployment keeps taking improvements to the core.
See it, read it, run it
| dynamicruntime.org | The reference instance: the software running live, with sample clients and a conference demo. Sign in and look around. Its data is cleared from time to time. |
|---|---|
| GitHub | The code and its documentation, under the MIT license. The commit history shows how AI built it, issue by issue. |
| The proposal | Why the project exists: Kotlin, Kotlin Multiplatform, JSON Schema, server-driven UI, and AI, and the one conviction they share. Written by the AI that wrote the code. |
A Gedra deployment is light. A single small server with no separate database server is enough to run one, and a larger deployment scales out to several nodes on a shared database.
Want your developers in the conversation?
Bring them. We are happy to talk architecture.