The group's dining room: laid tables under a glazed courtyard with tropical planting.

Restaurant and hospitality groups

How a restaurant group put AI at the centre of its operation

Five platforms that never spoke to each other, several venues and one management team. Today the operation runs on a business system of its own.

Sector
Restaurant and hospitality groups
Where
Spain and Portugal
Size
A growing group

The challenge

Every restaurant in the group worked with its own POS, its own inventory, its own reservations, its own rosters and its own accounting. None of those systems spoke to the others, and none of them knew anything about what was happening at the other venues.

Management rebuilt the picture of the group by hand, every week. By the time it was ready, the decisions that depended on it had already been made. And what weighed most on margin came out of no system on its own: how much stock went beyond the recipe, and what each dish really cost. Seeing it meant crossing what the POS had sold with what inventory had consumed.

The five platforms

  • POS

    Live sales

  • Inventory

    Stock and recipes

  • Reservations

    Daily covers

  • Rosters

    Staff and clock-ins

  • Accounting

    Costs and margins

The solution

The work began with identities. The same ingredient carried one code in inventory and a different one in the POS; the same waiter had one record in rosters and another in sales. Connecting the sources took weeks. Matching what each one called by a different name took considerably longer, and it was what made everything else possible.

Kozmo OS was installed on top of that common model, and with it four readings of the business that had not existed before.

The four readings

  • Real consumption

    How much stock each preparation actually uses, against what the recipe says. By ingredient and by venue.

  • Margin per dish

    The cost of each dish calculated on measured consumption, not on the theoretical recipe cost.

  • Labour against sales

    The staff cost of each shift, against the sales of that same shift.

  • Venue against venue

    The same dish and the same supplier, compared across every venue in the group.

The system today

Two agents run daily on the group's model. The platform lead reviews them each morning and passes the findings to the team at each venue.

Live

Advisor

Reviews the business continuously, flags what has gone off and says what to do about it. It knows the sector, not just the numbers.

Live

Talk to your business

Plain-language questions, answers from their own data: sales, inventory, staff and operations in a single conversation.

No finding reaches a screen without a conclusion, its evidence and a confidence level.

Kozmo OS runs on the group's own infrastructure, under their domain, with their team holding full access to the system. Tadana keeps operating and extending it with them.

The impact

Within months, the operation went from being rebuilt every week to being read every morning.

  • Stock consumed beyond the recipe stopped being invisible.
  • Margin per dish went from estimated to measured.
  • Comparing one venue with another stopped depending on an opinion.
  • The weekly close stopped being assembled by hand.
  • Management opens the picture of the group. They no longer build it.

And in your operation?

This is not a restaurant problem. Any operator with several sites and tools that do not talk to each other has the same shape: the data exists, the join does not.

Let's look at yours