Skip to content

Weather-Aware Activity Suggestion

Example prompt: "Every morning at 8am, look up today's local weather, then for each in-house guest in our Google Sheet, pick three to five activities from our 'Local Picks' tab tagged to match the day's weather and the guest's party type. Draft a short Gmail with the picks — leave as drafts, never auto-sent."

The Problem

A guest at breakfast asks 'what's good to do today if it's wet?' and the host has the same conversation five times before lunch. The walks we recommend on a sunny day are the wrong recommendation on a rainy one, the indoor picks for couples are the wrong picks for a family with toddlers, and the regulars we'd point the dog owner to never get mentioned to the walker without a dog. By the time the host gets to the keyboard the guests have already left for the day and the moment has passed.

How GloriaMundo Solves It

We build the morning briefing the host would write if they had time. The workflow reads the in-house list from the property sheet, checks the forecast for the property's town, and classifies the day against thresholds the host has set (not a guess) into one of four buckets — sunny and warm, mixed, rainy, cold and grey. For each guest, an LLM step picks three to five activities from the 'Local Picks' tab whose weather-condition tag and party-type tag match the day and the guest, preferring picks that also match a free-text booking note (anniversary, dog, toddlers, walking). The draft sits in Gmail for the host to read; nothing about a guest's stay goes out automatically. The Glass Box preview shows the day's classification, the in-house list, and one sample draft before the workflow does anything.

Example Workflow Steps

  1. Trigger (schedule): Every morning at 8am in the property's timezone.
  2. Step 1 (integration): Read the 'Bookings' tab for in-house guests where 'in_house_message_sent_today' is not today.
  3. Step 2 (integration): Read the 'Settings' tab for the property's town and the host's weather thresholds.
  4. Step 3 (web_search): Look up today's forecast for the property's town from a public source.
  5. Step 4 (llm): Classify the day against the host's thresholds into one of four buckets.
  6. Step 5 (integration): Read the 'Local Picks' tab and filter on weather-condition and party-type tags.
  7. Step 6 (llm): For each guest, pick three to five picks matching the day, the party type, and any free-text notes; draft a short Gmail and save as a Gmail draft.
  8. Step 7 (integration): Update each booking row's 'in_house_message_sent_today' so the workflow doesn't re-draft on a re-run.
  9. Step 8 (integration): Post a one-liner in #front-desk on Slack with the bucket, the in-house count, the draft count, and the first three draft links.

Integrations Used

  • Google Sheets — the 'Bookings' tab for the in-house list, the 'Settings' tab for thresholds, the 'Local Picks' tab as the curated recommendation source
  • Web search — today's forecast for the property's town from a public source (BBC Weather, Met Office, AccuWeather)
  • Gmail — the per-guest draft, saved for the host to send
  • Slack — the morning #front-desk summary so the host can sense-check the bucket and the drafts before sending

Who This Is For

Hosts at B&Bs, small hotels, and short-let portfolios where the conversation about what to do today happens over breakfast and the host's local knowledge is part of the offer. The picks live in a tab the host already maintains; the workflow just makes sure the right picks reach the right guest on the right kind of day.

Time & Cost Saved

The conversation at breakfast takes five to ten minutes per guest party and the host has it three or four times most mornings; on a busy weekend it can swallow the first hour of service. With four to six in-house parties on an average day, the drafts save the host thirty to forty-five minutes a day directly. The bigger gain is the guests who get a thoughtful note before they've left their room — they head out with a plan that fits the weather and their party, the host gets the time back, and the reviews mention 'the host's local recommendations' three months later.