Onboarding document for new employees

The problem

Canadian Heritage is a department of the Government of Canada responsible for policies and programmes relating to culture, arts, heritage, official languages, citizenship, multiculturalism, youth, and sport. Staff carry out file analysis as part of the department's work administering grants, funding, and policy decisions.

New employees joining the Programs Department team needed to consult multiple files and folders across different locations to find the information required to conduct this analysis. There was no single resource that brought all the guidance together, so new staff spent time searching for the right document before they could begin their work.

The solution

To address this, multiple existing guidance documents were consolidated into a single, unified resource. Microsoft Copilot, a generative AI tool, was used to identify and remove duplicate content across the source materials and to synthesise a more concise, cross-referenced version. This reduced redundancy and improved clarity across what had previously been separate, siloed documents.

A thorough manual review followed, checking each step of the analysis process for accuracy and completeness. Annotated screenshots of the system at key stages of the workflow were added to provide practical, step-by-step guidance.

The final output was delivered as both a formal guidance document and a PowerPoint presentation, providing new employees with accessible, structured learning materials.

The impact

This project introduced a single, centralised touchpoint for new employees, enabling them to complete their analysis in a single comprehensive document rather than consulting multiple files. This resulted in a clear gain in efficiency by reducing time spent searching for information and streamlining the overall analysis process.

Lessons learned

• A single consolidated resource was the main gain. The creation of one document that brings all relevant information together in one place was the most successful aspect of the project. It gave new employees a clear and accessible reference point and simplified the analysis process.
• Prompt design required significant iteration. Developing effective prompts to get the desired quality from the AI tool took several attempts. In some cases, Copilot rephrased or modified parts of existing processes beyond what was requested, introducing inconsistencies that required additional manual correction.
• Comparing outputs across multiple AI tools could improve efficiency. In future iterations, using multiple AI tools in parallel, such as Copilot alongside another generative AI assistant, to compare outputs and identify the most accurate result could reduce the number of prompt iterations required and improve overall productivity.