Three case studies, written around the decisions rather than the outcomes. The decisions are the part that transfers.
Deciding what an AI agent is allowed to do
An in-product AI agent that builds products with creators through conversation and answers questions from their own account data. I wrote the behavioural contract: what it may do, what it must ask permission for, and what it must refuse.
Building audience infrastructure for people who don't think in databases
A contact database with lists, segments and trigger-based email sequences, layered on top of a platform that previously sent one welcome email. The hard part was making a relational model legible to people who are not analysts.
Designing access for when payment and permission disagree
Paid membership communities with tiered access, a members-only resource library, events, paid sub-groups and a leaderboard. The hard part was access: one tier system holding a whole community, and rules that still make sense when a member has paid but has not been approved.