Are AI Agents About to Replace Traditional Hotel Software?

In a word, no. Websites were designed to entice humans to view, discover, consider, buy. They are heavy with images, HTML, JavaScript, etc. to entice us in and encourage us to dwell. APIs were invented for M2M, machine to machine, transactions. They are lean, measurable, secure. “Agentic AI” is still in its infancy. LLMs are still based on next-best-word, statistical, transformer-based Machine Learning engines. They have no internal real-world model. They don’t know up/down, before/after, time, distance, or especially numbers. They just mimic based on training data. “Agentic AI” is not a new AI advance; it’s just an LLM trying to orchestrate other LLMs. For heavy-duty, M2M, use deterministic RPA eg to migrate data from one PMS to another over a few hours. APIs can help if available, but RPA can read “green screens” (using Predictive AI) and move data from field A in system X to field B in system Y. An “agentic AI” agent might start ok, but fail when it reads Mr Stop, and then deletes entire guest profiles to speed up its task. PMS, POS, IBE, Spa etc systems aren’t going anywhere. They are based on solid deterministic programming. They already have APIs for inter-connectivity.

Agents will be everywhere, but behind the scenes.

Booking agents are a short-term solution; using AI to simulate human behavior is like asking for a faster horse. Agents that behave as people, visiting websites and simulating human behavior (clicking on links, filling forms) are an incredible technological feat, but are a poor solution to most real-word problems:

- Agents are very slow, because they have to replicate user paths built for humans

- They fail often, because every website they visit is different

The future will be MCP- and API-based, with agents interacting with those MCP/API. Using API-like calls is:

- Fast, because they are server to server

- Accurate, because they follow a common protocol

- Future-proof, because all AI companies have adopted this protocol

Crystal ball time. 5 to 10 years i think we will quite similar to where we are today. the slow pace of change will not change and traditional tech stacks will still be around. lack of motivation to change, contracts and low investment will hinder the change. but those coming in new to hospitality and can start form scratch will be able to take advantage of AI agents, but the old tech stack will again hinder this approach. we are just not setup for this, yet....

I doubt hotel software will disappear, and they will probably simply move down the stack. In that scenario, PMS, CRS, RMS, CRM, and other systems become structured data and transaction layers. The operational logic that today resides within each individual platform moves upward into an agent-driven orchestration layer.

The interface we interact with may no longer be the PMS dashboard or the RMS screen. It may instead be a conversational, post-keyboard, or task-based interface in which an agent coordinates multiple systems in the background, retrieves relevant data, and executes actions across them.

There is also a structural aspect specific to hospitality. Historically, this has been an extremely PMS-centric industry. An agent-mediated architecture could shift that center of gravity. Instead of being PMS-centric, the ecosystem could gradually become orchestrator-centric, in which the intelligence layer coordinates multiple systems and determines how they are used.

Such a shift could also reduce one of the long-standing issues of hotel technology: vendor lock-in. If orchestration happens at the agent level, switching individual systems becomes easier, and hotels may gain more flexibility, avoiding some of the rigid and sometimes unusual long-term contracts that have historically constrained the sector.

The idea that AI agents could interact with hotel software through graphical interfaces is certainly intriguing. This could shift interoperability from system-mediated integrations to agent-mediated orchestration. However, in hospitality technology the reality will likely be more evolutionary than disruptive.

Agentic AI can automate complex workflows across systems — for example handling booking processes, guest communication, or operational coordination — by orchestrating multiple platforms simultaneously. In that sense, intelligence will increasingly sit above the software layer, coordinating tasks across PMS, CRM, revenue management, and distribution systems.

But this does not mean traditional hotel software will disappear. Core systems remain the systems of record that manage structured data, transactions, and compliance. AI agents can only function effectively if they operate on top of reliable, well-governed data infrastructures.

There are also important concerns around privacy, security, and accountability. Agentic systems require deep access to sensitive data such as payment information, identity data, and guest behavior, which expands the potential risk surface if not carefully governed.

Looking ahead 5–10 years, hotel technology will likely evolve into a layered architecture: strong operational platforms at the core, with AI-driven orchestration layers on top. In other words, AI agents will augment hotel software — not replace it.

Exciting times indeed. The AI revolution will certainly shake up the hospitality tech landscape. Every software provider will have to re-establish its value in the value chain. Some tools will become obsolete, while new opportunities will emerge.

AI agents operating through a Graphical User Interface (GUI)—the visual interface of software with buttons, menus, and dashboards—could change how systems interact. Instead of relying purely on APIs, agents may operate software the same way humans do.

However, from what we see today, access to data remains the real power layer.

Even if agents can operate any system, they still depend on the ecosystems and data they can access. If you don’t have access to platforms like Google Search, Google Hotel Center, major distribution channels, or leading LLM providers, you miss critical parts of the market.

That’s also why we believe strongly in dynamic inventory and proprietary data layers at GauVendi—creating unique, structured product data that AI agents can actually understand and leverage.

Processes can be automated. Exclusive data and intelligence are much harder to commoditize.

The short answer is yes, but faster than most are prepared for.

I fully agree that interoperability is shifting from system-mediated to agent-mediated. This is not a subtle change. It is a structural one. When an agent can operate any software through its interface, every GUI becomes a functional endpoint. The practical path is hybrid: APIs where available, GUI interaction where not. Either way, the competitive value of traditional hotel software compresses to its data. Workflows, interfaces, and middleware become the agent’s domain.

The deeper shift is agent-to-agent commerce: autonomous agents coordinating across functions and organizations. Already emerging in adjacent industries. Hospitality is next.

I would push back on the 5–10 year framing. Within 2–3 years, traditional platforms will evolve into background data layers or be orchestrated by agents needing no permission to interoperate. Organizations with clean data move first; much of the industry will lag, deepening the competitive divide.

This will be driven by orchestration-layer companies and AI platforms, not legacy providers. If that intelligence belongs to a third party, hotels are not removing a dependency; they are adding a more powerful one. Those controlling their data retain autonomy. The rest will have a new landlord.

The bright agentic future won't be relying on human-like browsing. That method is not as reliable/scalable as the MCP/UCP approach and will likely only be used in the cases where MCP connection is not available, to close data gaps.

Humans will still be in the picture. Humans will still need data to look at (or talk to). Interfaces for humans will improve and become more user friendly.

The GUI as API framing is clever but it's solving the wrong problem.

The real constraint isn't interoperability, it's that hotels don't own their data in the first place. An agent navigating PMS, RMS, and OTA interfaces via GUI is still reading vendor data, stored in vendor schemas, behind vendor-controlled access. More sophisticated execution with the same ownership problem.

The hotels actually moving forward are building first party, warehouse-first architectures where PMS, booking engine, ESP, and consent records flow into a data layer they control. That's where agents can run on ground truth instead of scraped UI state.

Traditional software isn't going away, but its strategic value will compress toward transaction processing because the intelligence layer will live outside it. What matters is whether hotels own that layer or hand it to whoever sells them the agent platform.

Hotels that own their data can swap tools without losing anything. Hotels that don't will run more capable automation on the same fragmented stack and won't notice the problem until they try to change something. By then the gap is hard to close, and it has a real price: revenue left on the table and a lot of time spent firefighting instead of building.

It’s only natural that there will be highly democratized (ie. significantly cheaper), AI-driven hotel tech stacks in the near future. Just read the news and every week there are new single-point or multi-function tools emerging, now often integrated within an agentic orchestration platform.

While there will undoubtedly be an AI-first property management platform debut soon that offers a robust set of features for limited service / boutique hotels, the real obstacle may be a human one. How much appetite do our teams have for actual, hard change?

With this in mind, I envision a hybrid hotel tech future where innovative providers with established, global clientele rapidly roll out embedded AI tools. This would make for the most frictionless way to take advantage of it all over the next two years.

AI agents are unlikely to replace traditional hotel software entirely, but they will fundamentally change how it is used in the years ahead. Core systems for bookings, payments, finance and guest data will remain essential, because hospitality still depends on reliable systems of record and the business logic that keeps operations consistent and compliant. 

What will change is the layer above them. Increasingly, AI agents will play a greater role in this layer, orchestrating workflows across the tech stack and turning fragmented systems into something far more connected and responsive.

That is why APIs and MCP will become the new interface layer for hospitality. Instead of people clicking through dashboards, AI agents will increasingly act across systems through structured, machine-readable interfaces that can interpret intent and execute tasks. They will create an autonomous service layer around the hotel’s core system of record, and that is where a more agentic model of hospitality starts to emerge. This is how the industry gets closer to having truly autonomous hotels. Hotels that already operate on open platforms, naturally built to be compatible with these new ways of working, will find it much easier to transition their operations and take full advantage of autonomous operations.

This is one of the more consequential questions facing our industry right now.

AI agents will not replace hotel technology platforms. They will redefine how those platforms are used — and where value is created within them.

Core systems will persist as foundational data and process layers. What changes is the orchestration above them: agents navigating interfaces, automating workflows, and abstracting operational complexity that currently consumes disproportionate human attention.

This shifts competitive advantage up the stack. The differentiator is no longer the internal logic of any single platform, but the intelligence coordinating how multiple systems work together. That has meaningful implications for vendor strategy, procurement decisions, and how operators measure technology ROI.

However, agent-mediated interoperability does not reduce the need for robust data governance — it raises the stakes for it. Poorly structured data, inconsistent schemas, and security gaps become more consequential, not less, when an orchestration layer depends on them.

And hospitality's fundamental edge remains unchanged: judgement, empathy, and the human moments that build genuine loyalty are not automatable. The goal is agents absorbing operational complexity so people can focus precisely there.

The future is hybrid. Success belongs to those who treat data stewardship and operational rigour as strategic priorities — not afterthoughts.

One of the developments I have been most anticipating is the move from keyboard to dialogue as an input device for business technology. This is now emerging. That is exciting.

Ultimately, as Simone touches on, it will come down to structured data across traditional siloes. However, that challenge (conveniently) removed, the potential for useful information to be surfaced and interrogated through dialogue with technology is a game changer for human-computing interfaces in general.

Our strength as a species is the ability to speak. It is the swiftest and most efficient input-output pattern. And it is a core skillset in our industry.

I am of the view that we will move into a phase of even more value from technology through this shift.

AI agents won’t replace hotel software — they’ll sit on top of it.

Traditional systems (like PMS, RMS, payments) will still run the core operations. But instead of staff manually jumping between tools, AI agents will handle tasks across multiple systems automatically.

So the shift is:

From people using software

To software being coordinated by AI

In simple terms: Hotel systems become the “engines,” and AI becomes the “driver” that tells them what to do and connects everything together.

I believe that a 5- 10 year time frame is too far in the future to make a reliable prediction on how Al will evolve, especially as we still have limited knowledge on its capability to progress from ANI to AGI.  

Let's look at a few immediate changes we can expect. There will be a rush of Al agents, including those handling check-in, concierge, check-out, and revenue management, some supervised and some unsupervised. Then, we will see agents talking to agents to get things done. A flurry of agent activity, all using the hotel tech stack data as the base.  

Once guests start using agents to make direct bookings and negotiate deals with hotel agents, the agentic rush will slow down. That's when we in hospitality will start using Al-native solutions. Such solutions will help hotel staff greatly improve efficiency, leveraging Al to deliver exceptional guest experiences and boost profitability.  

There will definitely be changes in the interface layers; the number of functions will shrink, as they no longer need manual intervention. However, in some form or another, within the next 2 years, the hotel tech stack will continue to keep operational data in unified data layers to power Al. 

The right question, but wrong mechanism.

Computer-use agents that navigate GUIs are just a stopgap. They're brittle, slow, and outclassed by a consumer agent that calls a supplier's MCP server, which connects directly to the underlying API. Watching an agent click through a GUI dashboard is just painful.

The transition to APIs won't be simple, though. SaaS vendors are responding with competitive self-interest: HubSpot is charging tolls to reach data customers consider their own; Salesforce and SAP have invoked the "Balrog defense" (You shall not pass!), with SAP pushing it so far that Celonis sued in federal court just to maintain data access. This will cause customers to migrate to more open platforms.

A 5-10 year horizon seems overly broad. At today's pace, that's several AI generations. A more realistic window is 1-3 years, by which time OpenAI Frontier, Claude Cowork, and Gemini Enterprise will be the agentic platforms that forward-thinking applications integrate with. As Sam Altman put it: "Every company is an API company now, whether they want to be or not."

GUIs will remain alive and well on the web. In the agentic internet, computer-use agents navigating it will unwind quickly.

We are clearly in a platform shift, and this basically puts the score back to zero.

For many years, the model was simple: a human interacts with the UI, the UI executes business logic, and everything runs on top of systems of record. But now, “AI is eating UX”. A new layer is emerging: agents. They can understand context, decide actions, and use tools on our behalf.

However, this does not mean current systems will disappear. We will still need strong systems of record, deep vertical software, and solid data foundations. PMS, CRS or ERP will continue to play a key role, but their role is evolving. They are becoming tools that agents will use, instead of tools that humans directly operate.

The real shift is where the value sits. It moves from the software itself to the intelligence that orchestrates different systems and automates workflows end-to-end.

Also, this is good news from a practical perspective. We don’t need to rebuild everything. We can move fast by adding agents on top of the current tech stack, using existing systems, and start creating value quickly without full replacement.

AI agents are unlikely to replace traditional hotel software in the near term, but they will fundamentally reshape how value is created across the technology stack. Core systems—PMS, CRS, RMS, POS, CRM, finance, and operational platforms—will remain essential because they provide systems of record, transactional integrity, security, compliance, and data governance that businesses require. Hotels won’t run on agentic AI alone. But the interface, workflow, and increasingly the decision layer will change materially.

Agents will become a new orchestration fabric. In some cases, they will use APIs. In others, they will interact with existing interfaces where APIs are limited, fragmented, or slow to deploy. Software may still anchor the enterprise, but it may exert less gravity over where work happens and where users experience value. The center of activity moves from the application to the intelligence coordinating actions across applications.

This is how we think about it at Oracle: the future isn’t software disappearing—it’s becoming more composable, autonomous, and value-oriented. Over the next 5–10 years, the winning hotel technology environments will likely combine trusted enterprise platforms underneath with agentic intelligence on top, keeping applications as systems of record while intelligent orchestration become the system of execution.