Yes. Nib size, material (steel, gold), and flow tendency live in product meta. A customer asking for a wet medium under 200 EUR gets Pilot Custom 74, Pelikan M205, and TWSBI Eco picks from your real shelf with the actual prices and nib options in stock. Specific nib quirks (Lamy 2000 hooded, Sailor music nib) answer from a knowledge-base page rather than guessed.

If you maintain ink and paper notes in product meta or a knowledge-base page, yes. A customer asking whether Iroshizuku Kon-Peki feathers on Tomoe River 68 GSM gets the real tested answer. Generic chatbots invent compatibility; SleekAI quotes from your tested data. This matters because feathering is the most common return reason for fountain-pen customers buying their first journal.

Yes. Multibot plus display conditions lets each section run its own bot with its own prompt. The fountain-pen bot on /fountain-pens/ stays in nib-flow-ink territory. The planner bot on /planners/ focuses on layout and binding. The gift bot on /gifts/ builds curated bundles with cards and wrap. Each only sees products and pages relevant to its section.

Add your services pages to the knowledge base and the bot quotes grinds (cursive italic, stub, needlepoint), turnaround, and pricing. Refill compatibility for Lamy, Pilot, Pelikan, Caran d'Ache, and Schmidt fits answers from spec data. Customers asking which refill fits a specific pen get the right SKU rather than a Pinterest guess. Service revenue tends to lift once the bot starts surfacing the service at the right moment in chat.

Yes. Page count, ruling style (dot grid, line, blank, weekly, daily), paper weight, and binding live in product meta. A customer asking Hobonichi Techo Cousin versus Leuchtturm1917 Weekly gets a real comparison: A5 versus A6, daily versus weekly, Tomoe River versus 80 GSM. Bullet-journal starter kits pull notebook, pen, and ruler picks from one chat instead of three separate visits.

Yours. SleekAI is bring-your-own-key for OpenAI, Anthropic, Google, or OpenRouter. Most pen and planner questions answer fine on GPT-4o-mini or Claude Haiku at a fraction of a cent per chat. Reserve a larger model for nib-flow and ink-paper questions where context matters more. Usage stays in your provider dashboard, so spending is transparent and capped to your own limits.

Yes, when listed as products with release dates and allocation fields. The bot quotes street date, deposit policy, and waitlist position from your published pages. If you cap allocation per release, expose it as inventory and the bot tells customers when the run is full instead of inventing availability. Limited editions are a big draw for fountain-pen customers, and accurate waitlist answers prevent the worst kind of stationery disappointment.

Yes. Conversation logs surface re-stock signals, missing format coverage, and gift-bundle patterns. Repeated requests for an ink color you do not carry are a buying signal. Repeated nib-grind questions for a style you do not offer are a services signal. The log feeds buying meetings alongside POS data and informs which limited edition to commit to before the rep call.