Ninety-eight per cent of Australian operators now use AI in some form, and the number one application is menu and pricing optimisation, but the technology still needs a chef in the room.
For years, “data-driven” in hospitality meant reading last week’s sales report and having an opinion about it. That has changed quickly. Forecasting tools that read past sales, reservations and seasonal patterns are now inside ordinary Australian venues, and they are being pointed squarely at the menu.
Adoption is no longer early
Lightspeed’s State of the Hospitality Industry Report 2026, surveying more than 1,000 consumers and 300-plus venue decision-makers, found 98 per cent of operators use AI in some form, with more than one in four consumers considering AI essential for 2026. The leading use case is menu and pricing optimisation at 50 per cent, ahead of customer messaging (43 per cent), inventory (40 per cent) and scheduling (40 per cent).
Dependence is deepening. Twenty-seven per cent of Australian venues say they could not run the business without AI-powered tools, and 56 per cent save more than an hour a day using forecasting and optimisation tools that analyse past sales, reservations and seasonal trends to drive staffing, stock purchasing and menu planning. Venues using technology for business insights rose from just over three in ten in 2023 to more than half in 2025, according to Australian Bartender’s coverage of Lightspeed’s research.
Baseline adoption among smaller operators is solid too: 42 per cent of small-to-medium Australian hospitality businesses already use AI, according to Restaurant & Catering Australia’s coverage of the Australian AI Adoption Tracker. One Sydney operator reported AI handling 859 overflow calls in a month, converting around 200 into bookings and saving 20 to 40 staff hours a week, while computer-vision waste tools cut waste by up to 25 per cent within six months in international hotel pilots.
Where the forecasting pays
The hard results are in inventory and waste. London group Dishoom cut food waste 20 per cent using an AI inventory tool, as reported by The Age. Australian vendor analysis suggests AI inventory tools connected to POS reduce venue food waste 15 to 22 per cent within six months, worth $2,700 to $3,960 a month for a venue spending $18,000 a month on food, according to Calso’s 2026 analysis.
That figure is vendor-sourced and should be treated as directional, not as a guarantee.
We would also flag honestly that we have found no verified Australian venue-level data on dynamic pricing outcomes. Anyone selling you certainty on surge pricing in an Australian dining room is selling you a theory.
The cautionary tale worth remembering
Estelle in Northcote built an entire four-course menu with ChatGPT for the Melbourne Food & Wine Festival. Chef Scott Pickett said they used ChatGPT to propose the menu, timings, run sheet and wine matches using the restaurant’s archives. It also proposed a “Murray River red snapper”, a fish that does not exist, as reported by The Age.
That is the correct mental model. These tools are outstanding at pattern-finding across your own history and hopeless at knowing what is real. Forecasting belongs to the machine. Judgement stays with the chef.
A sensible sequence
- Clean your POS data first. Mis-mapped modifiers and stale item codes will poison every forecast.
- Start with demand forecasting for ordering and rostering, the highest-certainty return.
- Layer menu analysis second: item velocity, margin, cannibalisation and day-part behaviour.
- Keep a human veto on every menu, pricing and supplier decision.
- Measure against a pre-implementation baseline, or you will never know whether it worked.
Alto Cibum works with operators to turn POS and procurement data into decisions that actually change food cost and rosters, with the culinary judgement kept firmly in human hands. If you are unsure where to start, start with a conversation.