Two days on site at a boat manufacturer, and a plain answer about AI

Situation

The company's leadership wanted to know where AI could help the business and where it could not yet help. Leadership also wanted a clear view of what its IT team should work on next.

Solution

Before the visit, Jeff Hunt requested procedures, system diagrams, report samples and project lists. These were used to map the company's systems and mark where staff pull data into spreadsheets by hand. Jeff then spent 2 days on site interviewing leaders and staff and touring the factory. The written recommendations followed within 3 days.

For the IT team, the recommendations covered an annual roadmap of projects set with leadership, less custom software overall and AI tools to document existing spreadsheets and procedures. Projects to consider included a production scheduling optimizer and a way to read paper quality checklists into the system from a photo. A third would read vendor order confirmations and flag orders at risk of missing their date, and it was suggested as the first AI project.

Design decisions

The assessment advised against AI on the production floor for now. Predictive maintenance and quality prediction need years of production history, sensors on the equipment and data science staff. Camera-based defect detection is not yet reliable on curved white surfaces across many models built in small numbers. Their return is also not proven at the company's size.

AI was recommended for getting paper records into systems, documenting processes, email assistants for purchasing, monitoring industry and compliance news, and personal productivity. Adoption was to begin with interested staff and training led by IT.

Outcome

Leadership has a written project list for the IT team, a suggested first AI project and a clear statement of where AI does not yet justify its cost. The recommendations give the company a basis for setting an annual roadmap with its IT team.