Saving crane company hours of manual work

I created an AI agent that ranks the riskiest cars, cleans the data, and finds missed prestarts.

Comment

Inspiration

I observed supervisors manually reviewing spreadsheets for hours every day to monitor compliance while I worked for a crane company in WA. I wanted to use automation and artificial intelligence to address its messiness, slowness, and error-proneness.

What it does

Missed prestarts, roster inconsistencies, and non-compliance patterns are all detected by the feature. After that, it exports daily compliance + risk CSVs and ranks vehicles by risk over the previous 14 days. Supervisors can see immediately where they should take action.

How I built it

I extended the existing Data Cleaning Agent workshop framework and added a new feature notebook. Using Python, Pandas, and matplotlib, I built functions to clean messy datasets, compute risk scores, and generate visual dashboards.

Challenges I ran into

  • Handling messy real-world CSV exports from Numbers and Excel.

  • Deciding how to define and quantify “risk” in a meaningful way.

  • Balancing between making it run with or without an API key (fallback mode).

Accomplishments that we're proud of

  • Built a working AI agent on top of the workshop framework without breaking anything.

  • Automated compliance checks that used to take hours.

  • Delivered clear visual insights (trend line + top risk vehicles) in one pipeline.

What I learned

  • How to adapt a generic AI agent into a domain-specific use case.

  • The importance of clear data cleaning steps before any analytics.

  • That simple visuals can often communicate risk better than complex models.

What's next for Saving crane company hours of manual work

  • Embedding the pipeline into dashboards (Power BI / Tableau).

  • Adding real-time alerts when a vehicle crosses a “high-risk” threshold.

  • Scaling to multiple fleets and extending to other compliance checks (e.g., fatigue, speed).

Built With

Share this project:

Updates