Project progress

Lena Concierge: restaurant information for other AI assistants

Lena prepares restaurant profiles for external AI assistants: facts, sources and menus in one format. Progress and next steps as of 13 November 2026.

Illustration of a mountain village beneath snow-covered peaks and a low sun

An AI concierge for a resort can also be useful where guests already plan their trips: in other AI applications. On 12 November 2026, Lena Concierge added a restaurant profile lookup to its development version. It organises existing information about a single venue in a consistent structure. As of 13 November, this addition is not yet in the production version checked.

At a glance: location, cuisine, description, contacts, opening hours and available menus share one format. Sources and check dates can accompany them; missing information remains visible. This is a prepared data integration, not automatic table booking or confirmed admission to an AI app directory.

Why restaurant information should reach beyond one chat

A guest planning a stay in Saas-Fee or the Saas valley in Valais asks their usual assistant about a particular restaurant. To go beyond a general description, the application needs relevant local information tied to the right venue.

A name alone is not always enough. Similar names may describe different restaurants, while one venue may appear under both a short name and a longer official one. A menu without an identifiable date is also difficult to interpret.

Lena's new development component addresses that connection. It is designed to retrieve restaurant information already held, rather than relying only on fragments from a free-text knowledge search.

What was added on 12 November

A restaurant profile with consistent fields

For a venue it finds, the prepared lookup returns the same kinds of information: location, cuisine, description, opening hours, any recorded spending estimate, phone and email. A missing value is not replaced with invented information.

Each fact also has space for its source and last check date. Those may be absent too. A consistent structure therefore does not mean that every detail has been fully substantiated. It makes the gaps easier to see.

For a spending range, the explanation matters. An estimate based on a menu remains an estimate and does not become a binding restaurant offer.

Menus and seasonal details stay linked to the venue

Where a menu is held, the profile can include its sections, dishes, prices and sources. Where no dishes have been entered, the menu is absent. Conflicting menu dates should not be presented as one unambiguous issue date.

Mountain restaurants can also carry seasonal periods, lift access and separate restaurant and hot-kitchen hours. These describe information already held. They do not replace a check of actual operation or table availability.

An unclear name does not produce a guessed restaurant

If a request does not identify a single venue, the prepared function returns an alphabetical selection of the restaurant names held instead of a guessed profile.

This is a useful boundary for tourism information: clarification or deliberate selection is better than a detailed menu for the wrong establishment. The alphabetical list is not a ranking of quality or popularity.

What MCP contributes

The connection uses the Model Context Protocol, or MCP. The official MCP documentation describes an open standard that connects AI applications to external data and tools. Documentation checked on 13 November 2026.

For guests, what matters is the information accessible through it: a specifically requested restaurant profile with recognisable facts and gaps. A technical connection alone does not establish whether a particular application offers the function or how it presents the result.

The work on Lena therefore announces neither an AI provider partnership nor confirmed approval in a provider's directory. Public availability of the new profile function also needs checking after release.

Potential benefits for guests, hosts and destinations

Guests can get information about the right venue

The intended benefit is an answer clearly tied to the restaurant requested. Where a source or check date is available, a connected application can preserve that context.

Whether it actually does so, and whether the answer becomes clearer, needs testing in a real journey. No measured improvement is claimed here.

Hotels and holiday apartments need traceable recommendations

Hosts answer restaurant questions even when they do not run a restaurant themselves. A shared information base can help keep the venue's name, location, contact and menu together.

This could simplify information coordination, but it is not evidence of staff time already saved. Personal recommendations and the restaurant's own reply retain their roles.

Destinations can maintain information for reuse

In its Tourobs article of 20 August 2026, the Tourism Institute at HES-SO Valais-Wallis describes a digital data architecture with clear data objects, quality and interoperability as a destination responsibility. Source checked on 13 November.

Applied to Lena, a restaurant profile is one such object: a venue is connected to its information, which can then be used in different interfaces. This interpretation does not announce a new agreement with a destination.

What needs checking before use

The next steps are a verified release and acceptance testing in genuinely connected AI applications. Tests need to cover clear names, unknown venues, similar names and missing menus.

The visible answer also matters. Are sources and check dates retained? Is an estimate labelled as an estimate? Is an absent menu explained as unknown? Does the application offer an appropriate next step when the guest asks to book?

Before public release, published descriptions and legal texts must also match the actual functionality. Later evaluation should distinguish a profile retrieval from a helpful guest answer. Retrieval alone is neither a booking nor demonstrated business success.

Further reading

Frequently asked questions

What is a Lena restaurant profile?
It is a consistently structured response about one venue, containing available facts, sources, check dates and, where held, menu and seasonal information.
Has the new profile lookup been released?
As of 13 November 2026, it is prepared in the development version but is not in the production version checked. Availability needs confirmation after release.
Can another AI assistant book a table through it?
This profile lookup provides information. It does not create table reservations or confirm available seats.
Does every profile have a source and a current menu?
No. Sources, check dates and menus may be missing. The intended structure exposes these gaps rather than inventing information.
Does this automatically make Lena available in AI app directories?
No. A technical function and confirmed approval by an app directory are separate steps.