AI‑powered book discovery, as personal as your favorite bookseller.
Turn curiosity into connection. booklink helps readers find the right book—right now—with conversational, explainable recommendations that convert.
What booklink does
Conversational discovery that explains the “why,” with a drop‑in widget or API. Minutes to embed; measurable lift in CTR, add‑to‑cart, and time on page.
Ask by vibe
Readers speak naturally (e.g., “cozy heists,” “bittersweet sci‑fi”). We translate vibes into ranked results.
Explainable picks
Every recommendation includes rationale—tone, themes, comps—so it feels like a trusted bookseller.
Widget or API
Embed on search, category, PDPs, or email. Admin tools capture signals without extra work.
Solutions
Purpose‑built for retailers, publishers, and content platforms.
Retailers
Replace or augment site search; re‑rank category pages; lift add‑to‑cart with explainable recs.
Publishers
Surface midlist & backlist by intent, not keyword; turn catalogs into conversations.
Content Platforms
Increase session length & retention with the just‑right next book—on site, app, or chat.
How we prove value
Clear test design, privacy‑first practices, and operational readiness.
Ready to validate? Ask us for a tailored pilot plan.
World class recommendations with semantic understanding.
We encode books into semantically meaningful tokens so our model can understand, generate, and explain recommendations—guided by natural language and grounded in your catalog.
Architecture sketch
// Query → intents → semantic tokens → retrieve → rerank → rationale
user.query // “bittersweet sci‑fi with hope”
→ parse constraints (tone/themes/comps, series, age range)
→ LLM emits semantic tokens for likely books
→ retrieve candidates by token prefix + vector fallback from catalog
→ rerank with behavioral signals (clicks, CTR, availability, margin)
→ generate human‑tone rationale (guardrailed, cached)
→ return {items[], why[], links}
- Semantic understanding: Books map to short, meaningful tokens; similar titles share structure so the system generalizes well—even for new or long‑tail books.
- Guided LLM: The model “speaks” those tokens and follows natural‑language constraints (tone, comps, platform/age) to steer results.
- Hybrid ranking: Deterministic retrieval + signal‑aware reranking deliver control, speed, and measurable outcomes.
- Explainable: Every pick includes a concise “why”—in a trusted bookseller voice.
What this means for your catalog
Cold‑start resilience for new titles; a unified surface where readers ask by vibe, author, or comps and get explainable results.
Deployment & safety
Low‑latency lookups with caching; configurable business rules (availability, merchandising) in reranking; privacy‑first multi‑tenant design and short‑lived caches.
Why not pure RAG? RAG retrieves by text; semantic tokens let the model reason natively about items and follow constraints—then we keep it fast and measurable with deterministic retrieval and reranking.
Our Co‑Founders
Operators, builders, and book industry veterans bringing explainable AI to discovery.
Joel Silver
Cam Drew
Jordan Christensen
Like what you see?
Book a quick walkthrough or tell us about your use case.