Quartz 4

catalog-followon-ideas

4 min read

Catalog follow-on ideas

Brainstorm spinning out of the foundation built this week (see 2026-05-26 journal). Grouped by leverage and dependency. All build on the existing iPage adapter + Medusa importer + queue DB.

Quick wins on existing foundation (days each)

  • Weekly automated ingest with diff notification. Cron triggers just ingest <SFF> (or Indie Vault), posts a Discord summary: “+12 new books, 3 returned to filter, 5 dropped out, here’s the list.” Catches what’s worth attention without watching.
  • Cron stock-only sync. Already in medusa-inventory plan — lightweight refresh of just the stock column every few hours, independent of full ingest.
  • Restock alerts. Book in store drops below N on-hand at primary DC → Discord ping. Book transitions 0 → positive → “back in stock, restocking opportunity.” Just queries + thresholds against queue DB.
  • Duplicate edition detection. Same title (or close title fuzzy-match) + same author but different ISBN → likely hardcover/paperback pair. Surface so the catalog can decide consolidate or keep separate. (Earlier decision was “keep separate” — currently no tooling shows which products are paired.)
  • Catalog export for the accountant. CSV/PDF of every product with status, cost basis, last-sold-date, current inventory, BISAC. Pure DB query + format.
  • Pre-merge categorization preview. Before merging a change to categorizePath, run new rules against current queue DB and print diff (already built as a one-off script for 5 BISAC cleanup — formalize as just preview-rules).

Catalog intelligence on existing data (1-2 weeks each)

  • Recommendations engine. “If you liked X, try Y” using BISAC path overlap + co-purchase from order history (when there is any) + Ingram stock. Simple cosine over BISAC vectors gets 80% of value.
  • New-release alerts per customer. Watch wishlists, send “this released today” emails when pub_date matches. Needs customer wishlist infra in Medusa first.
  • Cover-change detection (deferred earlier). Hash + compare. Only urgent if placeholder→final-art rotation becomes a visible problem.
  • Going-out-of-print early warning. Book disappears from iPage detail entirely → flag in admin. Distinct from “OOS” — actually gone.
  • Phantom inventory check. Medusa says in-stock, iPage says zero everywhere → reconciliation report.

Add new discovery sources (weeks each, each independent)

  • Bookshop.org bestsellers scrape. Peer-indie signal, much more relevant than Amazon for an indie store’s stock decisions. Easy scrape.
  • Amazon “Books on TikTok” list. Already validated this scrape works at low volume. Resolves their Kindle-ASIN-heavy list to ISBN via title+author lookup against iPage.
  • Edelweiss+ buzz scores. Carrie probably has an account already. Surface upcoming-release buzz before books are even in BISAC.
  • Goodreads “Most Read This Week” by genre. Trend signal lagging Amazon/Bookshop but durable.

Operational AI tooling (weeks-to-months)

  • Back-office agent. Claude-Code-style agent that runs the weekly ingest, evaluates diffs against stated stocking criteria, and only pings when ambiguous. “I added 8 new BookTok books to Medusa as drafts; 3 were borderline (here’s why), please review.” Built on the MCP server from ipage-as-api.
  • Search-by-vibe. Customer-facing storefront feature. “Books with dragons, slow-burn romance, and a found-family arc” → BISAC + description embeddings + LLM filter. Distinguishes from B&N’s search. Real differentiator for a specialty store.
  • Customer-service chatbot that knows the catalog. “Do you have X? When can you get Y? What’s a good gift for someone who loves Sarah J Maas?” — wired to MCP, runs in Discord or on the storefront.

Strategic bets (months)

  • White-label the pipeline for other indies. The iPage adapter + Medusa importer + Discord bot is a real product for the long tail of bookshops that don’t have engineers. Probably 50-100 stores like Carrie’s nationally. Each pays $50-100/mo. Could be a real side-business with low ongoing cost.
  • Open-source the iPage Python client. Probably banned by Ingram’s TOS, but if not, real value to indie tech community. Could also be a recruiting/credibility play.
  • A curated-list-of-curated-lists. Aggregate Indie Vault + SFF + BookTok + Bookshop + Amazon BookTok + Goodreads into a single “ranked by indie-relevance” feed. Carrie acts on one feed instead of seven. Could be a product for other indies too.

Deliberately deprioritized

  • Custom AI book recommendations from scratch using embeddings — Goodreads/Bookshop already have decent ones, building a worse version isn’t worth it
  • Real-time iPage stock validation at checkout — the cadence problem isn’t urgent yet
  • Multi-source order routing (Bookshop vs Amazon vs Ingram) — only matters at a scale not yet reached

Suggested priority order

  1. Discord bot (captured in ipage-as-api) — operational lever, daily value
  2. Weekly ingest cron + Discord diff notification — turns the pipeline from manual to background
  3. Medusa inventory wiring (captured in medusa-inventory) — makes the storefront honest
  4. Bookshop.org bestseller scrape — adds a discovery source aligned with indie identity vs SFF top-N
  5. Recommendations engine — long-term differentiator

Graph View

Backlinks