What "consent-first AI" means in practice
The architectural decisions behind a personal AI that grows only from what you choose to share — enforced at the database, the prompt layer, and the audit log.
Essays, design notes, and the occasional confession from the small group building iSpirits Cloud in Sydney.
The AI-companion category is being sued, fined, and legislated for the exact design choices we refused to make. What "regulation-aligned by construction" looks like from the inside — and why it's a founder's decision, not a philosophy seminar.
The architectural decisions behind a personal AI that grows only from what you choose to share — enforced at the database, the prompt layer, and the audit log.
The five things to check before you trust an AI with your context — and how a consent-first iSpirit you raise compares to chatbots and journaling apps.
Most chatbots reset every session. Here's why that happens — and what an AI that remembers you does differently.
Customer journey maps were drawn for a slower century. What replaces them when you're running a tiny company on AI.
Conversation is the wrong unit of work. Inside the goal-driven runtime that walks each goal nightly and tells you what's next.
Every assistant reply names the memories it drew from. Inside the pgvector + HNSW pipeline that makes citations cheap and verifiable.
How iSpirit's Cloud Execution Center turns goals into completed work — on a credit budget you control, with results that save back to your journal.