01 · The position
The position
The productivity gains from AI are real. Left unchecked, they follow the same pattern as every previous wave of automation. Value concentrates at the top. Tools become proprietary. The people who need them most get priced out.
Studio exists to make that harder to do.
02 · Open source
Open source as structural choice
The AGPL-3.0 license is not a default. It is a deliberate mechanism against capture.
The kernel is a commons. Anyone can use it, extend it, build commercial products on it. If you modify Studio itself and run the modified version as a network service, you publish your changes. This prevents cloud providers from taking the engine, closing it, and selling it back.
This is not anti-commercial. Build whatever you want on top. Sell it. The license protects the foundation, not the products built on it.
03 · Domain-agnostic
Domain-agnostic by commitment
The engine has zero knowledge of what domain it operates in. It doesn't know what "code" or "transactions" or "medical records" mean. All domain semantics come from YAML configs.
This is not a feature, it is an architectural commitment that prevents any single use case from claiming ownership of the kernel. The engine belongs to everyone precisely because it serves no one domain specifically.
04 · Provider-agnostic
Provider-agnostic by commitment
Anthropic, OpenAI, or any other provider, swappable per agent, per stage, per pipeline. The orchestration layer depends on the work being done correctly, not on who does it.
No vendor lock-in. No preferred provider. The runtime is neutral infrastructure.
05 · Anti-capture
Anti-capture
Ideological drift rarely happens through open conflict. It happens through accumulation of "reasonable" decisions: a proprietary integration here, a vendor dependency there, a governance exception that becomes precedent.
Studio integrates mechanisms for deceleration, traceability, and veto. Any evolution that improves performance at the cost of increasing inequality is a regression, not progress.
06 · Governance
Governance
Decision-making power is held primarily by the people directly affected by the inequalities the project aims to reduce. Diversity is not symbolic, it is decisional.
The full governance model and foundational principles are documented in the charter.
07 · The goal
The goal
Reduce the cost of building software and automating complex work to the point where a single person can create and operate products that generate real value. Not by coding everything themselves, but by orchestrating AI agents with structural guarantees of quality.
The kernel is free. The agents are interchangeable. The configs belong to whoever writes them.