Technology · Stack
The first Renaix production standard.
Every layer below has operated in a real Renaix production system. It is a proven starting point, not a fixed bundle. Technology is selected according to each workflow's data, security, residency, integration, and operating requirements.
Stack overview
Four groups, matching the architecture.
Each group operates one part of the system.
Infrastructure
Operates the secured deployment control and the environment every layer runs in.
- Hetzner
- Docker
- Docker Compose
- Caddy
Workflow and data
Operates workflow orchestration, the governed data foundation, and the deterministic core.
- n8n
- PostgreSQL
- SQL
- Node.js
AI and agents
Operates selective AI, the domain agents, and the interfaces people reach them through.
- Anthropic
- Model Context Protocol
- TypeScript
- Zod
Operations and control
Operates evaluation, monitoring, and change control across the whole system.
- Automated tests
- Evaluation sets
- Git
- GitHub
- Deployment controls
- Monitoring
Detailed production standard
Every capability, its technology, and what Renaix is responsible for.
Infrastructure
Cloud infrastructure
Hetzner
Renaix responsibility
Provisioning, deployment environment, service operation
Containerized deployment
Docker, Docker Compose
Renaix responsibility
Reproducible services, isolation, versioned deployment recipes
TLS and routing
Caddy
Renaix responsibility
Secure endpoints, controlled public exposure
Workflow and data
Workflow orchestration
n8n
Renaix responsibility
Ingestion, extraction, matching, human review, scheduling, error workflows
Operational source of truth
PostgreSQL
Renaix responsibility
Schema design, reconciliation state, evidence, exceptions, role-scoped views
Deterministic processing
SQL, version-controlled Node.js
Renaix responsibility
Testable normalization, matching, validation, and export logic
AI and agents
AI extraction and reasoning
Anthropic Claude API
Renaix responsibility
Model integration, structured extraction, tool use, refusals, evaluation
Agent runtime
Node.js, TypeScript
Renaix responsibility
Domain-agent services, authentication, query logging, rate limits
Agent interoperability
Model Context Protocol
Renaix responsibility
One secure, versioned, revocable endpoint per domain agent
User interfaces
Claude and MCP-compatible clients
Renaix responsibility
Secure connection of approved users to role-scoped agents
Operations and control
Validation and testing
Zod, TypeScript, automated tests, evaluation sets
Renaix responsibility
Input and output validation, security and regression testing, acceptance gates
Source and deployment control
Git, GitHub
Renaix responsibility
Reviewable code, workflow definitions, deployment recipes, change history
The model layer is replaceable. The operating capability is not.
Hosted inference is provided by the selected model vendor, Anthropic in the systems Renaix runs today. Renaix owns and operates the integration, prompts, tools, evaluations, credentials, controls, and change lifecycle. The model layer can change without the business losing its governed data, workflows, or operating capability.