Salesforce's open-source cousin is having an AI midlife crisis
A metadata-driven platform rebranding itself as "ObjectStack" to chase the text-to-app dream.
What it does
Steedos is a low-code platform that lets you define enterprise apps through YAML metadata rather than boilerplate code. It auto-generates APIs, UIs, and workflows from *.object.yml and *.page.yml files, with a runtime that supports MongoDB, PostgreSQL, or even flat JSON files. Think Salesforce’s metadata model, but open-source and container-friendly.
The interesting bit The project is mid-pivot: the v2.x “Steedos Platform” is still supported, but all future development is heading toward “ObjectStack” — a three-layer stack (ObjectQL for schemas, ObjectOS for headless runtime, Object UI for React/Tailwind rendering). The bet is that AI can write the YAML faster than humans can drag components, making “text-to-app” the new drag-and-drop.
Key highlights
- Database-agnostic metadata layer: same YAML deploys to MongoDB, PostgreSQL, or JSON files
- Auto-generated GraphQL and REST APIs for every metadata-defined object
- Schema-driven React UI via Tailwind/Shadcn, with escape hatches for custom components
- Built-in RBAC down to field level, plus workflow automation through YAML or TypeScript triggers
- Docker one-liner (
steedos/steedos-community:3.0) ornpx create-steedos-appto scaffold
Caveats
- The ObjectStack rearchitecture is incomplete; the README warns v2.x is supported but future work is on the new stack
- The comparison table cheerfully marks Salesforce as “hard to extend” and “cloud only” — your mileage may vary
- 1,557 stars suggests a modest community; enterprise-grade claims vs. actual battle-testing is unclear from the README alone
Verdict Worth a Docker spin if you’re building internal CRUD-heavy tools and want a Salesforce-like metadata model without the licensing bill. Skip it if you need a mature, stable platform today — the pivot to ObjectStack means the ground is still moving.