Sample · The Stay Current Brief

Stay Current Proposal Site Brief

Answer a group inquiry with an interactive hotel proposal

Sample session takeaway for hotel sales teams and leaders • Updated September 28, 2026

What is new in AI

Four updates from the session, with a practical next step for hotels. The proposal site is the hands-on example; the other updates are worth watching or testing in their own right.

UpdateWhat it means for hotelsSuggested next step
Google Search Console AI reports [1]Worldwide rollout completed August 31. Reports show impressions of site URLs in generative AI features, not every mention of the hotel.Marketing: check which hotel pages appear.
GPT-Live-1 in the API [2]Released September 10 for developers building more natural voice experiences. Hotel uses could include inquiry handling or practice conversations.IT or vendor: evaluate a focused use case and handoff to staff.
Claude Docs and Slides [3]September 16 brought document, deck and design creation into conversations. This expands the workflow; file creation itself is not new.Sales: test one approved proposal or presentation.
ChatGPT Sites public beta [4]Builds and hosts interactive websites from instructions and source files. Availability and usage depend on the account and workspace.Sales: build one private proposal draft.

What the hotel example demonstrates

Lantern Court answers Morgan Ellis at Cedarline Advisory: 25 people, 20 rooms, two nights and a $28,000 hotel budget. Morgan needs private working time, local character and time for colleagues to get to know one another.

The proposal lets Morgan explore a three-day agenda, switch dinner formats and add an optional team activity. The agenda and estimate change together. The hotel source pack supplies the facts and pricing rules; the inquiry supplies the buyer’s priorities.

The task for your team Build one private draft, check it against approved source material, and discuss what still needs a human decision. Build time and visual results will vary.

Training example The hotel, client, people, prices and policies are fictional. The images are AI-generated concepts. Use approved property photography for real proposals.

The decisions still belong to the hotel

Good presentation can make an unsupported promise look convincing. Ask reviewers to check the offer and the behavior of the site, not just the writing.

Draft wordingJudgmentWhat the source supports
“The Foundry seats your 25 guests in a U-shape.”SupportedPublished U-shape capacity is 30. The Study holds only 12.
“Dinner in your own private restaurant.”ReviseA reserved section of Ember Table is not a private buyout.
“An easy walk to the French Quarter.”Remove or verifyThe source pack provides no walking distance or journey time.
“Your rooms are held until February 12.”ReviseThe quote expires February 12, 2027. Nothing is held; a signed agreement is required to reserve.

Before sharing with a client

  • Guests, rooms, nights, meeting dates and setup capacities match the inquiry.
  • Every option updates both the agenda and estimate. Room hire and standard AV are not charged again when already included.
  • Taxes, service charges, exclusions and offer expiry match approved source material.
  • Dietary requests, accessible-room details and weather arrangements are handled accurately; unresolved items are marked for confirmation.
  • Deposit, cancellation, attrition and payment terms are not invented or presented as agreed.
  • Photos accurately represent the hotel. Fictional training images remain clearly labeled.
  • The site works on a phone. Test every control, reset and print, including after revisions.
  • A reviewer approves the content and audience, then tests the link as the intended client.

Decide before the pilot

Choose an approved account, permitted client data, a reviewer and a person responsible for updates. Confirm brand rules and who may share externally. Sites is in public beta with plan limits; invited viewers may need to sign in. A private deployment is still a live site for its allowed audience. [4]

Our demo does not book rooms, collect payments, send email or update a CRM. Adding those functions, saved client selections or a proposal tracker requires additional setup and testing. A “tentative hold” in a tracker does not reserve inventory.

Build the sample proposal

Open Sites in ChatGPT, or choose Sites when starting a new conversation. Attach the two documents and all three images below before sending the prompt. A filename with “(1)” added by a download is the same training asset.

Training files

  • Lantern-Court-Hotel-Source-Pack.docx
  • Cedarline-Advisory-Retreat-Inquiry.docx
  • courtyard-hero.png • courtyard-king.png • foundry-meeting-room.png

Members download these training files in the member portal.

The prompt depends on these attachments. This brief alone does not contain all of the hotel facts, menus and policies needed to reproduce the example.

The full copy and paste prompt

Copy all five paragraphs below into the same message. This is the complete starter prompt from the session.

Create a polished, interactive proposal website from Lantern Court to Morgan Ellis at Cedarline Advisory, using the attached hotel source pack, client inquiry, and images.

Make it feel like a thoughtful hotel sales proposal written for this buyer. Lead with their priorities: private working time, local character, and helping colleagues get to know each other. Use the hotel facts and prices from the source pack. Label the hotel, client, and pricing as fictional training examples and the images as AI-generated concepts.

Include relevant guest rooms and meeting spaces, a proposed three-day agenda with a tab for each day, a clear estimate, and the next conversation with the salesperson. Use a warm boutique-hotel style with deep green, restrained gold accents, generous photography, and readable text. Make it work well on phones and larger screens.

Let the planner switch between the three-course dinner and casual shared dinner, and optionally add the Kitchen Table Challenge. Update both the day-two agenda and the full estimate when those choices change. Use the source pack’s service-charge and tax rules. Start with the three-course dinner and no activity. Include a reset and a way to print the selected proposal.

Do not invent facilities, availability guarantees, contract terms, reviews, distances, or extra fees. Clearly mark proposed arrangements that need confirmation. No live booking, payment, inquiry submission, or emails. Keep the Site private for my review.

Use the same approach at your hotel

Replace the fictional source pack, inquiry and images with approved materials. Update the names, options and style instructions to match. Remove the fictional labels only when the entire proposal uses verified real information. Start privately and complete the review above before sharing.

Practice one client revision

After the first build, paste this complete follow-up prompt into the same conversation.

Morgan says the team needs more time to connect. End the day-two working sessions at 2:30 p.m., serve the afternoon refreshment break then, and keep 3:00–5:00 p.m. open for informal conversations. Keep any selected Kitchen Table Challenge at 5:30 p.m. and dinner at 6:30 p.m. Update the day-two introduction to explain the change. Keep the full-day meeting package and all prices unchanged. Preserve the rest of the proposal and keep the site private.

Check March 17 after the edit. Then switch dinner formats and toggle the activity again: the revised afternoon should stay intact, and the price should only change when a priced option changes.

Check all four estimates

Dinner choiceKitchen Table ChallengeTotal with charges
Three-courseNot selected$25,676.40
Casual sharedNot selected$24,994.40
Three-courseSelected$26,869.90
Casual sharedSelected$26,187.90

Fictional USD totals for 25 guests, 40 room nights and 50 meeting-package person-days. Rooms use base × 1.14. Packages, reception, dinner and activity use base × 1.24 × 1.10. Transport, parking, extra drinks and other unquoted items are excluded. These are training assumptions, not actual New Orleans tax guidance.

Reinforce it in the next sales meeting

Have each seller bring one reviewed draft and identify an unsupported claim they corrected or a decision still needing confirmation. Track preparation and review time together. Compare similar inquiries and record client feedback; a small pilot cannot prove a conversion lift.

Other useful experiments: a wedding showcase, group arrival guide, meeting planner resource page, local itinerary or seasonal package page. Give each one approved sources and a clear owner.

Sources and session resources

Product information checked September 28, 2026. Availability can change. Hotel facts and all estimate checks come from the two fictional training documents listed above.

  1. [1] Google Search Central — Generative AI performance reports
  2. [2] OpenAI — GPT-Live-1 in the API
  3. [3] Claude release notes — September 16, 2026
  4. [4] Official OpenAI documentation — Sites

Questions: holly@hotelbschool.com

Get a Brief like this after every session.