Standard patterns fall short in highly regulated environments.

In most analytics systems execution is transient: a calculation runs, produces an output, and disappears. This is ideal for most data science but not regulated workflows where calculations must be documented, audited, replicated, and reused.

A different execution model

In trac, each calculation is defined upfront in an explicit metadata contract that drives execution and is preserved as the record of what was run.

Calculations are auditable, repeatable, and self-documenting.

Built in Tier-1 banking environments | Open-sourced via FINOS | Adopted by global systemically important banks

See Finnovate demo

How trac works

A contract based approach

Analytic execution is not a transient event — each calculation is the resolution of a governed durable contract.

Append-only data storage

Using standard cloud object storage, an append-only data model ensures that a complete immutable history of all analytic inputs is retained.

Virtualised model deployment

Model code remains in an external repository (e.g. GitHub) and is fetched dynamically at runtime, so every prior version of every model is a recoverable and executable asset.

Metadata-driven orchestration

Analytics is built around four governed object types each with their own metadata structure: data, models, flows and calculations.

Calculation contracts

Each calculation generates a calculation contract — a durable metadata record that both initiates execution and persists as the authoritative record of what was run.

Runtime resolution

Calculation contracts contain all information required for execution and are resolved at runtime, without reliance on mutable configuration or environment state.

trac architecture

Want the deeper technical view?

Read the calculation-as-contract white paper →

What intinsic governance means in practice

There is no deployment change risk becuse models, data, and process flows are immutable

All prior calculations can be reproduced and used as the start-point for further analysis

Governance documentation can be built from the contract metadata, not manually compiled

What trac enables

Transforming the regulated analytics operating model

By shifting the unit of control from environments to calculations, trac strengthens controls, improves analytical capability and reduces operational overhead.

The trac runtime can be installed independent of the platform. trac models can be run in your existing tools — from IDEs and notebooks to production environments — and will behave consistently across environments.

Learn more

By defining each calculation as an explicit contract, lineage, explainability and repeatability are intrinsic to execution — not enforced through separate tools or processes.

Learn more

Because every calculation is preserved as a reusable, queryable object, outputs can be combined, compared, and extended without rebuilding pipelines or reconstructing past runs.

Learn more

Transforming the operating model for: Model developers • Model users • Model risk managers • Auditors • Platform teams

Explore personas

Where trac sits

A governed execution layer for your analytics stack

trac sits alongside your existing tooling — introducing a governed execution layer that connects models, data, and compute into a controlled environment where every calculation is defined, executed, and preserved as a governed object.

Explore deployments
trac architecture