The OECD.AI Policy Navigator

Our policy navigator is a living repository from more than 80 jurisdictions and organisations. Use the filters to browse initiatives and find what you are looking for.

Codebridge


Added by:   OECD analyst
Added on:   16 Nov 2026
Updated by:   OECD analyst
Updated on:   16 Nov 2026

CodeBridge is an AI-powered tool created at Statistics Canada to translate statistical code into the open-source languages R and Python within a secure environment for protected and sensitive data. It also explains and documents code, supports customised prompts and enables staged conversion, such as from SAS to pseudocode and then to Python or R, in line with Statistics Canada’s AI guardrails.

Initiative overview

CodeBridge supports StatCan’s shift from proprietary applications to open-source software alternatives and aligns with modern data practices for using open-source tools. CodeBridge helps in the skills development of analysts and developers as it provides explanatory text for code and explains legacy “black box” code, facilitating the transition of statistical code to Python or R. 

CodeBridge provides fast, accurate conversions with clear explanations that help users learn as they go. It allows users to customise prompts, choose their preferred open-source packages (R or Python), and keep control of their code. It provides line-by-line explanations, so users understand what’s happening, like having a mentor built into their workflow. Code is processed in a protected-data environment, so data stays secure. The tool is offered to users in both English and French.

Two education surveys focused on secondary and post-secondary education finance underwent significant modernisation efforts since 2022. Both surveys were successfully transitioned to open-source platforms and have since completed multiple cycles using these new tools. Production in open-source tools has been completed for the 2023, 2024, and 2025 cycles. Since the initial migration process, additional tools were built in Python for digital collection tools, including automated mail-out processes and materials. 

The Blaise application, previously used for hard-copy questionnaires, became unavailable. Rather than allowing this disruption to impede progress, the teams proactively transitioned to R and Python for questionnaire development and data processing. Through the adoption of these flexible and powerful tools, the digital questionnaire processing environment was recreated using R, with early migration efforts supported via StatCan GitLab.

Results, outcomes and impacts

The migration effort began roughly three years ago on Azure Virtual Desktop (AVD) due to the absence of a designated Zone environment. This allowed the teams to establish processes and workflows in a cloud-based setting. Subsequently, work is transitioning to the Zone environment for enhanced stability and integration. One cycle has been completed using Python, which is used for data cleaning and verification. This open-source tool has been chosen to align with production partners in the Centre for Social Data Processing (CSDP), where migration is also underway.

Other relevant details

Challenges and lessons learned: As an increasing number of users from different subject matter areas access and use the tool, it has become clear that customisation and continuous improvement to meet all user needs will be a priority. The success of the tool to date is reliant on the underlying protected data environment and sustained funding.

About the policy initiative


Category:

  • AI policy initiatives, programmes and projects

Initiative type:

  • AI use cases/projects in the public sector

Status:

  • Active

Start Year:

  • 2026

OECD AI Principles:

—

Other relevant urls:

—