Our estimators want to use an AI tool to query our historical project files and PDFs to draft new project bids, but our shared drive is disorganized and full of outdated versions. How do we clean up this document hygiene before launching the tool?

Question

Tyler Smith · Accepted Answer

Before you allow an AI to query your historical project files, you must complete a basic document cleanup, or the tool will hallucinate outdated pricing and obsolete methods. You do not need to clean every single file in your drive, but you do need to define a single source of truth. Start by creating a separate folder specifically for approved reference materials. Move only the most accurate, closed, and successful project bids from the last two years into this folder. Ensure these files are saved in a consistent format with clear naming conventions. This step represents the Follow Thru conative instinct, which values systemization and order. If your team skips this and points the AI at a cluttered shared drive, the AI will pull outdated estimates from five years ago and ruin your margins. Once you have isolated the clean data, run a pilot with your lead estimator. Have them run test queries alongside their manual process to verify that the AI is retrieving the correct references. By establishing strict document boundaries first, you protect your company from operational errors and ensure your team can trust the outputs generated by the tool.