Typical operating reality today

Mdels are tied to environments, moving them implies reworking the code, reconfiguring the environment or containerising it all — creating friction in the route-to-live and between model developers, users and validators

The trac runtime separates models from environments so they behave consistently everywhere.

Self-describing models

A universal model API

Models declare their schema’s (inputs, outputs, parameters) and the trac runtime guarantees the model always receives appropriate inputs wherever it runs - handling data marshalling outside the model code, so developers can focus on model logic.

Explore the developer docs and start building trac models today.

Explore docs

Portable runtime

Code runs in any environment

The trac runtime is an open-source Python package that can be deployed in any modern IDE, notebook, or other execution environment, as well as in the trac platform.

It handles all interactions with the data and compute layer, so model code behaves consistently across environments — without rework.

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

See Finnovate demo

Automated deployment

A clean path from development to production

Models can be imported to trac from a code repository and deployed into executable workflows (Flows) using a seamless no-code interface.

No environment configuration, manual wiring, or code adjustement is required to go from development to production.

What this means

The same versioned model code can be used across development, validation and production teams — behaving consistently in every environment.

This removes rework, shortens the route-to-live, and allows teams to collaborate on a single, controlled asset.

See trac in practice

Explore how this would work in your data and analytics stack

Explore the Model API Book a consultation