Australia family AI travel planner: a proof-led planning test

2026-09-22 · Alfred Team · Travel notes

This proof test uses a Melbourne family travelling with two children, one old enough for full walking days and one who needs protected resets. The route is illustrative; live availability and prices must still be checked.

The traveller brief

• Two adults and two children aged 7 and 11.
• Eight days, Melbourne to Sydney, with a preference for fewer hotel changes.
• Moderate pace, one major anchor per day, a midday reset and flexible dinners.
• One rail transfer and one weather-sensitive outdoor day.

The route spine

Use Melbourne as the first base for three nights, then protect the Melbourne–Sydney transfer as its own decision rather than squeezing another attraction around it. Finish with four Sydney nights so the family can recover from the move and keep the final days editable.

Named decisions and fallbacks

• Melbourne day 1: keep the first afternoon close to the accommodation base; fallback is an earlier dinner and reset if arrival fatigue is high.
• Melbourne day 2: place one named central-city anchor before lunch; fallback is a shorter nearby route, not a second cross-city transfer.
• Transfer day: protect the station-to-accommodation sequence and do not add a timed attraction before check-in.
• Sydney day 1: use one harbour-side anchor with an indoor fallback if weather changes.
• Sydney day 2: choose one family-interest block and leave the afternoon open for energy or weather.

What the proof must show

The finished Alfred flow should capture the family profile, dates, pace and route constraints; produce a structured itinerary; let the traveller edit a day; and make sharing with the other adult obvious. The plan should retain the route rationale after an edit rather than becoming an unstructured list.