HOW FIDELITY INVESTMENTS IS SEEKING VALUE THROUGH CALM FOR STANDARDIZED ARCHITECTURE OUTPUTS

Turning architecture from static, team-specific documentation into reusable, machine-readable, and continuously governed design assets.

The Challenge

Architecture documentation and artifacts are often created in different formats and stored across multiple repositories. Diagrams may be maintained in presentation files or drawing tools. Policies typically have a defined home, but access may be restricted, while derived standards may circulate as documents that are difficult to interpret and apply consistently. The result is not a shortage of architecture information, but a shortage of consistent, connected, machine-readable architecture documentation.

For Fidelity, this fragmentation made architecture harder to reuse, validate, govern, and keep current. Static diagrams could communicate vision to people, but they could not be reliably queried by tools, checked automatically against controls, or consumed consistently by AI-assisted engineering workflows.

We needed a way to standardize architecture documentation. The value is in capturing knowledge, design intent, dependent relationships, controls, and architecture decisions in a form that allows people, pipelines, governance platforms, and AI agents to run throughout a delivery lifecycle.

The CALM BASED Solution 

Fidelity is partnering with the FINOS Common Architecture Language Model (CALM) community as a standardized notation at the center of its Architecture-as-Code approach. CALM provides a structured, version-controlled representation of architecture. Fidelity extends that foundation with governed knowledge, reusable assets, design-time validation, and integration with existing governance systems.

The Fidelity approach treats an architecture design as a durable living source of design intent rather than a one-time deliverable. Within a CALM JSON, nodes describe what exists, relationships describe how elements connect, controls describe what is required, and metadata provides context, references, and layout. Human-readable Markdown complements the model with business requirements, design context, security expectations, and usage guidance.

Fidelity organizes each architecture package into four complementary layers:

Layer

Purpose Typical Content

1. CALM

Machine-readable architecture Nodes, relationships, interfaces, required controls, dependencies, and diagrams generated from the model.
2. Documentation Human and machine context Business requirements, usage guidance, design flows, security expectations, and direct references to CALM elements.
3. Validators Automatable assurance Schema validation and policy checks against controls
4. Human review Accountable decision-making Architectural review of the model and narrative to confirm correctness, context, exceptions, and fitness for purpose.
5. Package Long term storage Package the current state for historical review before any future changes

 

industry value

Fidelity’s use of CALM illustrates how an open architecture specification can become a practical foundation for enterprise standardization without forcing everyone into an identical internal operating model.

  • A common language creates portability across tools and reduces dependence on proprietary diagram formats.
  • Machine-readable models allow architecture validation, visualization, documentation, and AI-assisted workflows to build on the same source.
  • Reusable patterns and building blocks allow common architecture challenges to be solved once and improved collaboratively.
  • A shared standard can improve interoperability among architects, engineers, risk teams, controls, and delivery platforms.

fidelity value

For Fidelity, the value is a more unified architecture practice that standardizes outputs while preserving accountable architectural decision-making.

  • Consistent architecture packages: models, diagrams, narrative, decisions, controls, and evidence can follow a repeatable structure.
  • Shift Left Architecture: Controls can be evaluated in design time, in the pull request, or pipeline rather than only after design completion.
  • Reusable enterprise knowledge: consistent patterns, standards, and platform nodes can be applied across business units.
  • Improved traceability: architecture assets are version-controlled and can reference governed standards, decisions, catalog identifiers, and validation evidence.
  • AI-ready context: structured architecture and governed Markdown provide richer input for design assistance, specification generation, code generation, and conformance checking.

Questions

If you have questions, email us at info@finos.org.

 

Fidelity Vertical Case Study Graphic (1)

Learn More & Get Involved

Visit the FINOS CALM project site to explore the specification, tooling, working-group materials, and contribution paths. Organizations can participate by testing CALM in design-time architecture workflows, contributing reusable patterns or tooling, and sharing implementation lessons with the Architecture as Code community.

calm.finos.org