First post and intro

I started answering real consumer problems before deciding what to build. Something unexpected happened.

I kept finding the same pattern behind completely different problems: a cancelled subscription that still gets billed, a refund marked as processed that cannot be traced, a contractor invoice that no longer matches the quote, or a payment request that doesn't clearly match the agreed work.

So instead of building another general AI assistant, I'm experimenting with very small problem-specific protocols.

Each one starts with one concrete situation and helps a person reconstruct the records, identify what matches, what doesn't, what is missing, and what they should clarify before making their own decision.

I deliberately don't let the AI decide who is right.

I've now turned six of these situations into 2-page protocols under SPIRIGI.

What I'm trying to understand next is: which everyday document or billing problem repeatedly costs people far more time than it should?

I'm especially interested in cases where different systems, companies or departments appear to hold different versions of what happened.