In most software organizations, system architecture lives in a graveyard of outdated slide decks, Confluence pages, and static whiteboard diagrams. The moment a system goes into production and engineers make their first hotfix, the diagram on the wiki is already wrong.
CALM (Common Architecture Language Model), an open-source project hosted by FINOS (the Fintech Open Source Foundation), aims to solve this problem by turning software architecture from static pictures into living, machine-readable code.
What is FINOS CALM?
CALM is a standardized, JSON Schema-based specification for defining software architectures in a format that both humans and machines can read.
Just as Infrastructure as Code (IaC) tools like Terraform replaced manual server configuration with version-controlled scripts, CALM brings the "Architecture as Code" (AasC) paradigm to system design. Instead of drawing boxes and arrows by hand, architects and engineers define their systems using structured data that describes:
- Nodes: The building blocks of a system (microservices, databases, APIs, external systems).
- Relationships & Interfaces: How those components connect and communicate with one another.
- Flows: End-to-end business processes and data pathways across components.
- Controls: Security, compliance, and operational requirements attached directly to the architecture.
Because CALM uses JSON Schema under the hood, organizations can define reusable architectural patterns (standards) and automatically validate whether specific team architectures conform to those patterns.
Why is CALM an Important Project?
Especially in highly regulated industries like financial services—where thousands of interconnected systems must meet strict security, resilience, and regulatory requirements—traditional architecture practices break down at scale. Here is why CALM is a game-changer:
1. It Eliminates "Architecture Drift"
When architecture lives in a drawing tool, it is disconnected from the codebase. With CALM, architecture definitions live right alongside application code in Git repositories. They go through standard Pull Request (PR) reviews, version control, and CI/CD pipelines, ensuring the documented architecture reflects what is actually running in production.
2. Automated Governance and Compliance
In large enterprises, getting an architecture approved often means weeks of manual review boards and spreadsheets. CALM allows organizations to codify their architectural standards and regulatory controls into schemas. Using the CALM CLI, teams can automatically validate their system design in seconds inside a CI pipeline—catching compliance or security gaps before code is even deployed.
3. Generate Diagrams Instead of Drawing Them
Instead of spending hours nudging boxes in Visio or Lucidchart, teams can treat diagrams as a byproduct of the model. Because CALM captures the underlying structure and relationships in a standardized format, tools can automatically generate C4 views, sequence diagrams, and documentation straight from the single source of truth.
4. Makes Architecture Ready for AI
Large Language Models (LLMs) struggle to accurately reason about messy, inconsistent diagrams across hundreds of slide decks. Because CALM represents architecture as structured, semantic data, AI tools can easily parse, query, and generate it. This unlocks powerful capabilities like:
- Automated Blast-Radius Analysis: Asking an AI assistant, "If this database goes down, which business flows and customer-facing services are impacted?"
- Architecture Discovery: Generating or updating CALM models directly from source code and infrastructure configs.
5. Vendor-Neutral Industry Standard
Historically, every enterprise architecture tool used its own proprietary format, locking companies into specific vendors. By developing CALM openly under FINOS with contributions from major financial institutions and technology firms, the industry gets a shared, vendor-neutral language—complete with open-source CLI tools, VS Code extensions, and visualization integrations.
The Bottom Line
Software delivery has automated testing, security scanning, and infrastructure provisioning—yet architecture has remained stubbornly manual. FINOS CALM bridges that final gap. By treating architecture as code, it turns system design from a bureaucratic bottleneck into an automated, verifiable, and AI-ready part of the modern engineering workflow.