"Don't clone the reference. Find its one good idea and generalize it past what it could do."
Two References, Two Borrowed Ideas
Bonfire doesn't invent from nothing — it stands on two reference apps and takes exactly one idea from each. Fretastic already renders one scale across two instruments (fretboard ↔ keyboard). That proves the 'one model, multiple views' idea is real and useful. Capo already analyzes audio into chords, key, and tempo. That proves Tier-A analysis is solved. Bonfire's job isn't to rebuild either — it's to take their proven ideas further than they go.
The Delta, Precisely
Fretastic Bonfire
--------- -------
views 2 (fb, kb) N (fb, kb, staff, tab, ...)
model static scale living song / solo
explanation none Pippa Sidekick
analysis input manual Capo-style audio -> chords/key (adapter)
Three moves: 2 views become N (the polymorphic-view architecture generalizes Fretastic's two), a static scale becomes a living song (the model is now a whole piece with chords and notes over time, not a fixed scale shape), and Pippa is added to talk about the model. That's the delta in one breath: 2→N, scale→song, +Pippa.
Start Where the Reference Stops
Capo is the clearest example of the principle. Capo analyzes audio and stops at labeling — here are the chords, here's the key, good luck. Bonfire treats that labeling as an input adapter and starts exactly where Capo stops: now explain why the progression works, simplify it to the player's level, coach the improv, convert it between instrument views. The reference's finish line is Bonfire's starting line. That's how you build on a mature tool without competing with it — you consume its output as your input.