Oracle's no-cost OPERA Cloud AI changes the ROI question for hotels

The company is putting AI inside the property management system that many hotels already use. Oracle OPERA Cloud Assistant adds AI capabilities to the front desk, revenue management and configuration workflows within OPERA Cloud. OPERA Cloud customers worldwide can use it at no additional cost. Wyndham is among the brands already using the assistant, and it operates more than 2,100 properties on OPERA Cloud.

That account comes from Robert Firpo-Cappiello, vice president of communications at HFTP, whose report on the show ran in Hospitality Net and Hotel-Online. He wrote that what he heard most often at the show were questions, uncertainty and a need for expert guidance, including: What is the return on all this investment? He also wrote that Oracle's no-cost release changes the ROI math for those properties. When the feature comes at no additional cost, the purchase stops being the hard part. Choosing the workflow, the owner and the measure of success is what's left.

For a hotel CIO or a management company's operations lead, that's a different job from running a vendor search. It's also arriving on a deadline. Otel, which sells an AI layer that sits above a hotel's existing systems, published its strategy guide in late October for general managers and revenue managers who have been told to put an AI line in their 2027 budget and don't know where to start.

AI built into the PMS, POS and phones

Oracle was not the only vendor taking this route. Agilysys built its AI strategy, guestsense.ai, on embedding features in its core systems rather than selling a separate add-on. That strategy rests on four pillars: multi-modal user experiences, hyper-personalization, agentic process automation and revenue intelligence. Under them, the company has rolled out more than 30 AI-powered features and modules. These include AI-native features in Agilysys PMS and enhancements to Book, InfoGenesis POS and Eatec. Two more modules are still in development: Revenue Intelligence and a next-generation central reservations system.

Canary Technologies designed its guest management platform with AI at the center, offering AI guest messaging, AI Voice for calls and upselling tools.

Oracle designed its own OPERA Cloud Assistant for staff. Its features include AI-assisted room assignments, AI-generated rate descriptions, multilingual translation and plain-language answers when staff ask about operations. Oracle also provides AI Agent Studio, which is used to build agents around industry workflows. Oracle has built about 50 agents so far, though it has acknowledged that many hotels still do not have a clear idea of which agents they want.

The decision that matters is which of the available features gets pointed at which task.

Behind the desk first

Its subtitle describes technology as a support for people and a way to run hotels better. Hotel Dive's Lara Ewen covered the findings. Among the hospitality workers surveyed, 90% said AI lets them get through routine tasks more quickly. Respondents ranked operational efficiency first for value at 38%, roughly 10 points above revenue growth at 28%.

Results for workloads and frontline jobs looked very different. Some 84% of respondents reported that AI had eased their workloads, but only 15% saw any gain in frontline work. The report reads that gap as a sign that bringing AI into guest-facing roles may be among the biggest openings the industry has. It also argues that AI does the most good by supporting hotel staff, and that it should not replace essential hospitality labor.

Of those surveyed, 59 percent wanted people, not technology, to keep leading the front-desk welcome and check-in. The heaviest AI users held that view most strongly.

Nsabimana's column also addresses what AI should do for staff. He describes AI in a support role, organizing guest information, spotting preferences, translating and flagging unresolved issues. He argues that some of the time it frees should go back to guests, not into more work. That's a recommendation, not a measured result.

The rulebook is running behind

Nearly all hoteliers in the Mews survey, 98 percent, reported using AI during the prior six months, but 41 percent were operating without a formal AI policy. Trust in AI was nearly twice as common where a written policy existed. Otel also reports figures from its own sessions this year, which it describes as a show of hands, not a formal survey. More than 90% of attendees used tools like ChatGPT, Copilot or Claude at work, fewer than 5% worked at a hotel with a defined AI strategy, and more than 95% planned to have one by 2027.

Spending plans point the same way. A 2026 Canary Technologies survey of more than 400 hotel technology decision-makers, cited in Otel's announcement, found 85% plan to put at least 5% of their IT budget into AI this year.

Both Otel and Nsabimana offer concrete ways to handle that. Otel's guide runs a four-step method: find the repetitive work, choose one use case, pilot it for four weeks alongside current practice, then decide from the evidence. Nsabimana adds escalation rules. Staff review AI outputs before anything guest-facing is automated. Repeat failures, strong emotion, and anything involving safety or money go to a named person, and staff keep a practical override.

One way to draw the line is by how serious the problem is. Nsabimana points to a 2026 study published in the Journal of Consumer Behaviour that ran four experiments using service robots in unmanned smart hotels. When failures were minor, customers forgave an apology from the robot alone more readily than a joint apology from a human and a robot. When failures were serious, having a human involved produced more forgiveness. The authors recommend human intervention for serious failures, such as a guest being unable to check in.

Who keeps the guest data

The panel that opened the show's AI discussion, moderated by PwC's Jeanelle Johnson, spent much of its time on distribution. Benu Aggarwal, founder and president of Milestone, expects more bookings to come through online travel agencies. Even so, she said, hotels can still own it directly by building a relationship with the guest. Priya Chandnani, EVP of commercial strategy at Sage Hospitality Group, split the transaction from the stay: whoever takes the booking gets the data, while the hotel owns the stay and the communication around it.

Peter Chambers, Montage International's EVP and CIO, said Montage wants AI for speed and scale, not to replace people. He called AI possibly the industry's greatest opportunity, while noting it could also let entirely new businesses step in between hotels and their guests. His remedy was the hotel's story. In his view, what a property offers beyond its room rate is what will get it found by AI. Chandnani suggested selling curated experiences, such as a spa getaway, instead of just rooms.

Aggarwal's first fix connects this section to the previous one. She said data consistency comes first, so large language models can find the hotel. She described a Milestone project in which all of a brand's data, gathered over years, went into a newly built layer designed to understand and optimize that information.

The next markers

According to Firpo-Cappiello, hotel companies want evidence that AI is paying off, but the metrics for measuring that are still being developed. Oracle has about 50 agents built already, yet it has acknowledged that many hotels still do not have a clear idea of which agents they want.