How recommendations get made.

We maintain a small, hand-curated database of sweet spots — redemption opportunities where the points-to-cash ratio is meaningfully better than the market norm — and a separate ledger of active transfer bonuses between major loyalty programs.

When you have an open trip and balances that match a sweet spot, we compute a headline dollar gap (what you’d save vs. paying cash or vs. an obvious alternative redemption) and surface the single best-fit option. We score for fit, not for novelty.

We use language models to draft the human-readable explanation of a recommendation, but the recommendation itself is selected by deterministic rules from verified data — not generated. Every dollar figure cites the underlying sweet-spot record.

This page will get more concrete as we ship.