Oracle's Tanya Pratt: Cloud was the foundation, AI is the decoration

We started where we always do, asking Tanya Pratt, global vice president of OPERA Cloud strategy and product management at Oracle Hospitality, how she'd describe the company to someone who didn't know it. The question half-dissolved on its own, because everyone knows Oracle, MICROS and the Fidelio history, so we followed her own story instead. She has spent 32 years in hospitality, and like a lot of the best people in it she started at the front desk.

She came to Oracle in 2019 after 24 years on the operator side at Fairmont. Halfway through that career a CIO from outside the industry, an aerospace guy, pulled her out of operations and into IT, on the logic that he could teach a hospitality person technology faster than he could teach a technologist hospitality. What made her say yes to Oracle was the scope. At a single brand you stay true to one brand's DNA; at Oracle she went from 100 luxury hotels to working across economy, midscale, resorts and casinos, and as she put it the playground got a thousand times bigger. Casinos were new to her, she'd never been to one before, lost $180 on her first visit, and never gambled again.

Why staffing is the one she won't answer

Tanya is an expert in Hospitality Net's World Panel, and we asked about it. She said she doesn't submit her view on each Viewpoint that we sent out, because some sit too close for her comfort, and one she chose not to answer was about staffing. Not because it's awkward for Oracle, but the opposite: it's too personal. The moment anyone raises staffing it goes straight to her heart, because she was one of those people on the front desk, and she wouldn't have her job now without having started there.

Her worry is specific: when she hears about frontline staff being displaced, she thinks about where the leaders of the future come from. As she put it, "you're not just born into leadership." Most good leaders got there because of everything they learned working their way up from the floor, and she's not sure what it does to the next generation if you remove the rung everyone used to start on.

Cloud was the foundation, AI is the decoration

We asked what the next big step is after the move to cloud. Cloud, she said, was talked about for 10 years before it happened, held back by managers who didn't want to lose control of their own database and their own queries. What actually pushed it over was the pandemic. Hotels reopened and staff didn't come back, and owners looked at the server in the basement "that once a day, somebody would press a button on" and asked what they were still doing with it.

On top of that, AI changes how Oracle builds. They spent years getting the cloud foundation solid, scalable, and secure, and now, she says, AI lets them "decorate it." The house is built; the features go up far faster because the team works differently now. She described the old way as sequential, product writes it down, hands it to engineering, code comes back in two and a half months, everyone looks at it and sometimes sends it back to the start. Now they sit together every day. She calls it "sandcastles": kids show up at the beach not knowing each other, one has a bucket, one has a shovel, and a few hours later there's this thing. For a customer that used to file an enhancement request and hear "see you in 18 months," that's the real change.

She made another point. Hospitality has never been on the cutting edge of technology; some other industry was always ahead. AI and cloud, she reckons, have evened the playing field, and she'd now put hospitality ahead of utilities or comms or pharma.

Fifty agents, and nobody asking for them

What frustrates her is that the appetite hasn't caught up to what's possible. Oracle has built about 50 agents already, from understanding the pain points that have sat unsolved in the industry for decades, and every time she shows them the reaction is the same: oh my god, this could be huge. And yet she's "still so shocked," she said, that nobody walks in with a list of 50 agents they want. It ends up being a conversation, a lot of prompting on her side.

She goes out of her way to talk to the right people. A CIO lies awake at night thinking about something different from a front office manager, and she goes to the ones in the trenches, because they're who Oracle actually services. Her test for any agent is whether it solves a real business problem. If a GM looks at it and says this does me no good, there's no point building it. That discipline, she said, is probably the number one thing they practice.

Asked for concrete examples, she gave four. A support agent, so instead of calling support to ask how to undo a checkout, you just ask it and it walks you through. Room assignment that reads the guest profile and the reservation, including unstructured text, so a travel agent's note like "guest on honeymoon, please upgrade" gets acted on instead of waiting for someone to run an arrival report and read every comment. A fully agentic check-in, where you type "check in Tanya P" and it authorizes the card, assigns the room, takes the ID, and frees the agent to actually welcome the guest. And one for the sales team: standing on the floor, say "hold three hundred rooms in the grand ballroom next August" and it blocks the inventory, no calling the office, no running back. Sounds simple, she said, but for people who know, that one's "life-changing." Some are in production now; more land in August.

Search yes, booking she's not sure

On distribution, Oracle is looking at MCP, and for search and booking inside a tool like ChatGPT. We pushed on the part nobody's settled, whether guests will actually book that way. Tanya said she's a skeptic on the transaction. Personally, she likes AI for search, but for the booking itself she still prefers the normal display, and she'll admit she's been in the industry too long to be the easy convert.

Even as a researcher she finds the AI answer unsatisfying, because when it hands her the top five hotels, her first thought is: what about the ones you didn't show me. A long shopping list she can scan in a glance; a chatbot makes her keep asking and asking. She thinks the whole industry is about to learn, fast, how users actually behave between those two modes, and she's not pretending to know the answer. Her point is that the most disruptive thing about AI is its unpredictability. When Oracle sold cloud, the ROI was clear: someone else handles security and infrastructure, total cost of ownership drops, staff focus on the guest. With AI, she said, "we still don't know the ROI." Is it lifting revenue, are satisfaction scores up, can a hotel reduce staff because check-in is faster. Nobody can show the results yet that become the playbook everyone follows.

What she wants to hear next year

We asked where Oracle will stand in a year, and Tanya didn't give a metric. What she wants is for a hotel to tell her is that Oracle made a real difference to the business, helped the team, helped the guests, helped them run the place better. Because the moment that happens, the customer starts saying give me more, I can think of more, and that's the loop she wants, to crowdsource the innovation rather than hand it down. Fraud prevention is high on her own list, and the longer goal is to ship as much as possible out of the box, included in OPERA Cloud, so hotels can add their own sprinkles on the cupcake without carrying the whole investment themselves.