Do You Think You're Ready for A2A Commerce?
Originally published in the Hotel Yearbook 2026 Technology Edition.
The industry still likes to think of agentic AI as a distant, chatbot-adjacent phenomenon. It isn't. The real disruption is not travelers talking to AI assistants — it is machines negotiating directly with machines, and a hospitality infrastructure built entirely around human browsing behavior that is nowhere near ready for it.
You wake up on a Tuesday morning and casually say to your AI assistant: Find me a beachfront hotel in San Diego for three nights in July. Quiet at night, strong Wi-Fi, healthy breakfast, walkable area. A few seconds later, you receive a notification: "I booked this for you."
No browser tabs opened. No scrolling through reviews. No comparing rates across six different websites. Just machine-to-machine collaboration — and not just collaboration, but negotiation. This sounds futuristic, but it is already happening.
What the article covers
- This is bigger than chatbots. The difference between an AI assistant and an AI agent is action: an assistant tells you which hotel is best, an agent goes and books it. The real transformation is AI systems connecting and negotiating directly with other AI systems and supplier infrastructure.
- Machines talking to machines. MCP helps AI systems securely access data sources such as hotel rates and offers; A2A lets AI systems communicate and coordinate with one another. Together, AI shifts from conversational to operational — and the distribution structure begins to change.
- Hotels optimized for humans. AI doesn't care. AI agents don't respond to drone footage or urgency messaging. They evaluate structured data, machine-readable policies, trusted inventory access, interoperability, and real-time accuracy.
- A once-in-30-years reset. The orchestration layer of travel distribution is being rebuilt, and whoever controls orchestration controls demand flow. Hotels that fail to build agent-accessible infrastructure risk handing that layer to intermediaries once again.
- The hard truth: hospitality is not ready. PMSs, CRSs, booking engines, loyalty platforms, and revenue tools remain deeply fragmented, with inconsistent APIs and room attributes that autonomous systems can't reliably interpret.
The question is no longer simply "How do hotels market to travelers?" It is: "How do hotels become understandable, trusted, and accessible to machines acting on behalf of travelers?" The new customer may no longer be the traveler — it may be the system representing them.