Answer
The strongest pilot candidates have numbers attached already. Accounts payable is a common starting point because most finance teams can already state how many invoices move through the process each month and roughly how long manual matching and filing takes per invoice, which makes the before-and-after comparison concrete instead of anecdotal. Shipping paperwork is another good candidate when customer disputes or chargebacks trace back to missing or hard-to-find proof of delivery, since resolving those faster has a direct dollar impact leadership already cares about.
Before starting, define what success looks like in specific terms, average time to retrieve a document, percentage of documents indexed without manual correction, or reduction in filing and refiling labor, so the pilot produces evidence rather than an opinion. The pilot also doubles as a real test of the integration approach with the core IBM i application, which matters because problems with index accuracy or workflow routing are far cheaper to fix on one process than to discover after a portfolio-wide rollout is already underway.