Larkspur Outfitters, a fictional US multichannel seller with a small software line, needs a defensible tax-evidence recommendation. Build six reusable records covering nexus, product mapping, cited research, certificates/AP/notices, integration/reconciliation, and exposure/audit support. Compare deterministic controls with AI-assisted extraction, drafting and classification, and preserve accountable source and acceptance decisions. Real-state materials are sourced and dated; fictional jurisdiction rules are clearly separated.
The planned15-hour supplied-output route needs no paid product, model account, code execution or real client information. Larger Python/model workshops are optional extensions. Six module Decision Labs and a strategic capstone test choices after changed evidence. The final lesson links fallible graded practice feedback and revision; that exercise is not an additional certificate requirement or tax/legal clearance. Independent reviews are required before release; this author packet claims no approval.