Choose AI methods for forecasting, visual inspection, drafting, and routing.

Instruction: Recommend an approach and evaluation for each workflow, explaining when a non-AI method is sufficient.

Context:

A fictional manufacturer has four needs: forecast next month’s demand by product using historical sales and permitted planning data; flag visible surface defects from product photographs; draft support replies from approved articles for human review; and route tickets using an authoritative table mapping customer tier and issue category to a queue. Demand history and labeled inspection images exist, but their quality has not been assessed. The routing table changes monthly.

Updated

Official answer available

Read the opening below, then unlock the full answer and practical guidance.

For demand, I'd compare a predictive forecasting model with a simple seasonal baseline. I'd evaluate future periods using only information available at each forecast date and check errors by product. I'd also ask how stockouts affect recorded sales, because sales may understate actual demand...

Upgrade to view official answer

Related Questions