Start by comparing the workflow and the cost of keeping it working. Custom software is useful when important requirements cannot be met sensibly through an existing product or configuration. It should earn its extra responsibility.
Compare three routes - Buy: an established product already covers the main workflow. Check pricing at your expected usage, data export, permissions and integrations. - Configure or connect: the product is close, but needs forms, automation, reporting or connections to other systems. Check who owns those additions and maintains them. - Build: a distinctive workflow, user experience or integration constraint justifies a product you control. Budget for operation and continued change as well as development.
Write down the five requirements that matter most. For each option, record what is supported, what needs a workaround and what cannot be done. Include licensing, migration, support and exit costs. A lower build price is not automatically a lower total cost.
AI is a separate choice. A predictable rule, validated form or ordinary search can be more appropriate than a model-generated answer. Ask what the AI improves and how you would measure that improvement.
Use Software Discovery when a product direction needs a credible delivery scope. If you need an independent recommendation before deciding to build, discuss that decision first.
Bring to the discussion: the current tool, one example workflow, the most expensive limitation and expected usage. You do not need to upload confidential records.