Questions, answered.

Most rooms publish in under an hour. Paste a menu URL or your spec sheet, pick an archetype, set a logo and palette, and print the placard. The concierge runs the moment you publish.

Stripe collects your card at checkout, but we do not charge it during the trial. After seven days, your card is charged $199 monthly or $1,990 annually, depending on the plan you picked. A reminder email lands 24 hours before we bill.

Yes. Cancel from your portal. You keep access until the end of the billing period. No retention maze. We also offer a 30-day money-back guarantee — email support and we refund the last charge in full.

The concierge doesn't ring orders today. It prepares the guest and hands the call to your floor team, who still take the order in Toast, Square, Resy, OpenTable — whatever you already run. POS-side order push for Toast and Square is on the public roadmap for late 2026; the underlying data export means you can pipe recommendations into POS reports right now.

Each location runs its own brand config — its own menu, voice, and live URL — under one billing account. Group dashboards roll metrics up across rooms. If you operate more than 5 rooms, ask us about multi-unit pricing; the per-room subscription drops materially at scale.

Yes. Every guest conversation, menu item, and brand asset stays yours. Export to CSV any time from the portal. On cancellation, you can request a full export and full deletion — both happen within 30 days. We are not in the business of holding your data hostage.

Yes. The guest microsite ships at our hosted URL by default, but you can point a subdomain (sip.yourdomain.com) at it via DNS in a few minutes. Full white-label — your colors, your fonts, no FluteOS chrome — is included on every plan. Reseller white-label with your own checkout is a custom arrangement; reach out.

The concierge recommends from your list — if a wine, beer, or spirit isn't on your menu, it can't recommend it. We don't sell or dispense alcohol; we don't take payment for drinks. The order itself is taken by your licensed staff at the bar or table, in the same POS and venue you already operate, under the same liquor license. We never collect ID and never confirm an order — those steps stay with your team.

Guests must affirm they are of legal drinking age before entering the concierge experience — the prompt is built in and configurable to your jurisdiction's threshold (21 in the US, 18 in most of the EU, etc.). The affirmation is logged with the session for your records. Underage guests are routed to a non-alcohol-only mode if you choose to enable one.

Recommendations are constrained to your imported list — bottle, beer, or cocktail must exist in your catalog to be poured. The concierge cannot recommend an SKU it hasn't seen. Tasting notes and pairing language ride on top of the catalog; we ground the model in your menu first and never let it free-recommend by name. If a guest asks for something off-list, the concierge says so and pivots to the closest in-house pour.

Two layers. First, a system prompt tuned per archetype that holds the concierge to recommendation, pairing, and brand-voice tasks — no medical advice, no political commentary, no refunds, no comparing your room to a competitor's. Second, a runtime safety filter that blocks unsafe categories regardless of how a guest phrases the ask. Custom guardrails (no off-menu mentions, no spirit recommendations after 11pm, etc.) plug in via your portal.

We currently run on the latest Claude (Anthropic) models for text and OpenAI's GPT-5 family for some niche tasks. Voice mode uses ElevenLabs. The model selection is invisible to your guest and we benchmark monthly against new releases — you get whichever model lands the best pairing-accuracy score on our internal eval, with full backward compatibility on prompts and outputs.

No. Your menu, your brand voice, your guest conversations — none of it is used to train any model, ours or our providers'. We have data-processing addenda with Anthropic, OpenAI, and ElevenLabs that disable training on your traffic. Aggregated, fully de-identified telemetry (latency, error rates) is the only thing we retain at the platform level.

No. Beverage Concierge sells standalone for $199/mo or $1,990/yr. If you already run FluteOS on Growth or above, it's included at no extra cost.

Yes. A one-click upgrade is rolling out. In the meantime, ping sales and we'll switch plans in one working day, with credit applied for unused time on your concierge subscription.