Spanish hotel technology company Mirai has begun rebuilding its booking engine to support conversational, agent-driven travel transactions, a shift that ties one of Europe's largest independent hotel platform operators into the same protocol race now reshaping how AI assistants route reservations to suppliers.

The Madrid-based firm, which manages direct-channel technology and distribution services for hotel clients across Spain, Portugal and Latin America, frames the rebuild as a response to the moment travelers stop typing queries into search boxes and start negotiating itineraries with chatbots and AI agents. The booking engine — historically a structured form-based checkout — is being re-architected to accept conversational inputs, agent-issued requests and the structured intents behind them.

What does "agent-driven" actually change?

For hotels working with Mirai, the distinction is operational, not cosmetic. A conventional booking engine requires a traveler to land on a property page, select dates, pick a rate, enter payment details and confirm. An agent-driven flow hands that sequence to a software intermediary — increasingly a large language model acting on the traveler's behalf — which queries availability, compares rates, applies loyalty terms and returns a bookable result. The seller's loss if that flow is not supported is revenue routed to a competitor's system or to an OTA whose inventory the model already recognizes.

Where Mirai sits in the stack

Mirai's portfolio spans direct booking technology, demand generation and distribution services; its hotel clients include large Spanish chains and a long tail of independent properties across Europe and the Americas. Re-engineering the booking engine is the highest-leverage move in that stack because the engine is the conversion point where lookers become bookers and where commission leakage to OTAs is either captured or lost. If the rebuilt engine can accept an agent-mediated request, the property keeps the direct booking. If it cannot, the request defaults to whichever platform has trained the model to recognize its inventory.

The competitive pressure

Mirai's move lands against a broader pattern in which Booking Holdings, Expedia Group, Airbnb and a growing list of startups are publishing APIs and protocols built specifically for AI agents. Google has been routing increasing share of travel queries through AI Overviews and Gemini-based assistants, while independent projects push standardized agent-to-supplier communication. For Mirai and the hotels it serves, the strategic question is no longer whether AI agents will touch their booking path — that transition is already visible across referral traffic — but whether direct-channel economics hold up as the path grows more mediated.

What suppliers should watch

Hotels connected to the rebuilt engine will see whether conversational flows convert at parity with form-based checkouts, and how attribute-rich the agent-recognized inventory turns out to be. The commercial stakes are concrete: every percentage point of conversion lost to a non-direct channel represents permanent margin eroded at a moment when rate integrity is already under pressure from rate-parity clauses and OTA commissions that routinely sit in the 15–25% range. Operators in Mirai's footprint that move first on agent-readable inventory will set the benchmark for the rest of the Spanish-speaking independent hotel sector.

The deployment timeline and the first wave of hotels testing the rebuilt engine will determine whether Mirai converts a defensive technology refresh into a distribution edge — or simply keeps pace with platforms that have already started buying traffic through AI assistants.