Infor's David Poprawka: hotels will build their own AI, and the real rivals are Google and Anthropic.

Infor is one of the bigger enterprise software companies, and its hospitality arm makes the core systems a lot of hotels run on: the property management system, the revenue management system, the point of sale, and more. We started by asking how Infor uses AI internally, and David Poprawka, an innovation strategist at Infor, said it's a different beast to coordinate at Infor's size, with corporate and global restrictions front of mind. Internally they're using Claude to model use cases for agentic AI, which led straight to what they came to HITEC to show: an orchestration and intelligence layer they call Infor Portico Hospitality AI.

The name does a lot of work, and David clearly enjoys it. A portico is a threshold, the entrance into something larger, and that's the whole pitch: a doorway clients step through, out of the chaos on the other side, where they're bombarded with AI and can't tell what's useful from what's noise. Past that threshold sits a workspace where Infor and the client make sense of agentic AI in their own context. He was careful about what it is and isn't: not another chatbot, not a single agent, but an agentic platform that sits above the core systems you already have, POS, PMS, RMS, and connects through both open API and MCP.

Build the workspace, not the app

David kept pressing a distinction: between building a fixed product and building a place to build. Most vendors, he said, ship a generic concierge app and then try to bend it to fit each hotel, which works for five or 10 use cases and falls apart at scale. Infor went the other way. They spent the last couple of years building the orchestration layer itself, a space where a hotel designs the agents that fit its own operation, whether it does that with Infor's help or by itself. And because Infor Portico Hospitality AI sits above the core systems and reaches them through API and MCP, it can run on top of other vendors' systems too, not just Infor's.

You do it through conversation. A chatbot works with you to figure out your use cases: a concierge agent, or something for sales and events, or an agent that drafts function sheets, or one that helps a large convention centre qualify the 2,000 leads it can field at once. David didn't want to spend time on the replace-humans debate, not because he dismisses it, but because he takes it as obvious that nobody's aiming for that, so saying it out loud adds nothing. Infor Portico Hospitality AI gives clients a creative workspace to build the agents that make sense for their specific question, at the time they have it.

The agent they built during the storm

For a concrete example, David reached for something they had built the day before, on the spot. Storms and heavy rain across the south-east US were delaying and cancelling flights during the show. The idea came from a colleague, Leonardo: build an agent that creates a predictive housekeeping priority list based on which guests are likely to be delayed. They connected Infor Portico Hospitality AI to Flight Radar and to a second data source, and David's aside was that the second one burned through millions of tokens. Tokens are the units AI models charge by, so a connection that pulls in a constant flood of data can get expensive fast, and that was a real lesson they can now pass to clients: be careful what you connect and what data you let feed a request, because the cost can outrun the value of the use case.

As a user, this is how it works. You're the front office manager at a hotel next to a convention centre, it's the busiest stretch of the week, thousands of arrivals, and you can't give everyone the full treatment. The traditional route is to look up your VIPs and loyalty guests and prioritise those. But if the system can see a guest is physically on a delayed plane and won't arrive until six in the evening, there's no sense having housekeeping rush that room and letting the welcome amenity sit there going stale. You redirect that effort to five other guests you know will arrive early, because you have their flight information too, and you create a great arrival for them instead. Maybe you turn a guest loyal to another brand through that one human encounter.

A connectivity layer, not a new database

We put it to him directly: Infor Portico Hospitality AI is really a connectivity layer to the LLMs, not a data layer, because the data never moves, it stays in the core systems. David confirmed it. The core solutions stay where they are; Infor Portico Hospitality AI doesn't replace them and isn't trying to. It reaches into the data sitting unused inside those separate, disconnected systems and brings it together on one layer, where it can actually be put to work. They don't connect to a whole system, they select only the specific API calls that serve a given function. The LLMs are tools used to model that data and even build a predictive layer on top, all in service of letting a client build the workflow they actually want, for the value it creates rather than workflow for its own sake.

From vendor to educator

Clients mostly don't understand MCP, agents, or the new technology that's suddenly possible, and David said Infor's role is shifting from vendor to educator, which changes the whole sales cycle. Instead of a one-stop-shop product that either fits or doesn't, they now do hyper-personalised consulting first, to find the real problem.

One example showed what that gets you. A large tribal casino, which he didn't want named, comps all its rooms, meaning it gives them away free to its gamblers rather than selling them, which creates a roughly $35 million annual revenue displacement. For years it couldn't connect its loyalty system to its PMS to handle high-spenders in real time. Connecting those two and putting a generative tool through Infor Portico Hospitality AI on the problem, Infor built what they call a Whale Finder, a real-time VIP arrivals list, against that $35 million opportunity. It won't replicate exactly anywhere else, which is the whole point of Infor Portico Hospitality AI.

David takes the logic to its edge. He thinks the industry is moving past off-the-shelf solutions, because hotels can increasingly build their own agents and may lean on vendors less. The competition, he says, isn't other hospitality vendors anymore; it's Google, OpenAI, and Anthropic, because nothing stops those players from offering the service direct.

What he actually wants AI to be for

David's own story came up here. He started his career at the Waldorf Astoria, on Hilton's management development programme, a fast track to GM that he called the hardest and best school of his life. He worked under Eric Long, the longtime GM of the Waldorf Astoria in New York, a man who turned down SVP and EVP roles because he called the hotel his life's work and wouldn't abandon her.

Long had one rule: never let a guest leave unsatisfied. He gave David every power and every budget to make a guest happy, no matter how long it took or what it cost, but if he saw a guest walk out unhappy, there would be a problem. David said that period broke him down, nights awake, crying in the back office, and made him who he is. And this is where it connects to AI: the work that exhausted him then was the manual part, pulling lists, chasing references, researching. That's exactly what AI can do now.

So his view of AI isn't efficiency. He'll even say it might make us dumber. For him, the point is that it frees people to concentrate on the thing that makes them special, his own willingness to go above and beyond, the soul and passion he doesn't want the industry to lose. He worries the biggest risk of AI is that hospitality loses that human core entirely. If AI can instead hand a young receptionist with a heart of gold back the time to create a real guest experience, then for him that's the only use case worth standing for.