Projects / ai-labs / dididecks-ai / models/ Models
Data + content models in dididecks-ai (as-observed-from-filesystem) - What the client-sites’ filesystem-as-database actually looks like — so the collaborator designing the remote DB stops guessing at shapes.
Auth surface — Identity, Session, Token, Organization, Membership, AuthEvent - The auth tables that gate every deck — already SQLite-shaped (astro:db on libSQL, Turso remote), unlike the .md-on-disk models.
Company — portfolio companies + brand-asset metadata - One markdown file per portfolio company with colocated brand assets; the discriminator against Person is the absence of role_class.
Deck / Variant / Slot — the deck-OS registry - The deck → variant → slot spine — src/data/{decks,slides}.ts read at build time; the filesystem is source of truth for slot existence.
Engagement telemetry — PageView + Action append-only logs - Append-only PageView + Action logs in the auth astro:db — the foundation for ‘who read what, for how long’ reporting back to founders.
Firm — VC firm / backer firm metadata - One firm.md per firm-anchored directory — present in calmstorm and chroma’s multi-investor layout, absent from humain’s flat one.
Person — team members, advisors, portfolio CEOs - One markdown file per person plus colocated headshot; role_class discriminates vc-team, advisor, portco-ceo, and the rest.
Slide audit registry — per-slot dual-surface review ratings - The only runtime-mutable model on disk: one JSON per client holding per-(variant, slot) ratings, written by /api/slide-rank.
Substantiation corpus — source decks, intake material, raw assets - The gitignored <client>/corpus/ layer — founder PDFs, page-burst PNGs, memos, raw assets. The DB stores pointers to it, never the content.