Every hotel tech provider we talk to has an AI roadmap now. And a common question comes with it: if AI can help engineering teams move faster, do you still need to invest in a separate integration and data layer? Or can AI just help you build the integrations yourself?

Here's the honest answer: AI can help you build integrations. But it can also help us build integrations. And, integration and data are our products.

The question isn't whether AI makes it easier to write integration code. It does. The question is where you want your engineering team's time and resources going.

If your team uses AI to build a point-to-point integration, your engineers still have to review the output, test it, catch errors and hallucinations, handle edge cases, get through certification, and maintain the integration once it's in production. AI can accelerate parts of that work, but the work still belongs to someone.

A recent industry survey backs this up: Sonar's 2026 State of Code Developer Survey found that AI already accounts for 42% of committed code, yet 96% of developers don't fully trust AI-generated code, and only 48% always verify it before committing

Meanwhile, Hapi's engineering team is using AI too. The difference is that we're applying it to the problem we've already spent years solving: connecting hospitality systems and making their data usable at scale.

AI is only as good as the data model underneath it

“Garbage in, garbage out” applies just as much to AI as it did to every system before it. Whatever you're building (a guest messaging assistant, a personalization engine, or anything else that needs to understand a guest) needs clean, structured, real-time data to work with.

AI can absolutely help a team write code faster. What it doesn't give you automatically is a normalized hospitality data model that works consistently across every system.

Before AI, or anyone, can move fast across PMS systems, someone still has to answer a deceptively hard question: what does "guest," "reservation," or "folio" actually mean across OPERA, Infor, Mews and whatever comes next?

Each system stores that information differently, updates it on its own schedule, and has its own quirks and edge cases. Reconciling all of that into one consistent structure isn't just a coding problem. It requires years of hospitality-specific knowledge, decisions, testing, and maintenance.

We've spent years building that normalized layer, PMS by PMS, edge case by edge case.

AI can help us move faster too. But it doesn't erase the foundation we've already built.

The work doesn't end when the integration goes live

There's another part of the “we can just build it with AI” argument that gets overlooked: building the integration is only the beginning.

Someone still has to validate what the AI produced. Someone has to test it against real systems and real-world edge cases. Someone has to work through certification with the PMS vendor. And someone has to maintain it when the vendor changes something six months later.

That's especially important in hospitality, where integrations aren't static pieces of code sitting quietly in the background.

Systems change. APIs change. Data structures evolve. Vendors release new functionality. Properties configure systems differently. And when something breaks in production, someone has to figure out why.

And if integration isn't your product, that means your engineering team is spending time maintaining infrastructure instead of building the features that differentiate you in the market.

Relationships still matter

AI can help with the technical work of building an integration, but it doesn't replace the relationships behind that work.

Hapi has longstanding relationships with PMS vendors and their engineering teams, built through years of working together on integrations. We know the certification processes, understand how different vendors approach their systems, and have established points of contact when questions or issues come up.

Those relationships are part of what makes integration work at scale possible. When you've been working with the same vendors for years, you bring context and trust to every new integration and every change that comes along.

AI accelerates fragmentation, not just development

As more companies adopt AI, the pace of new tools, features, and data sources is going to accelerate for everyone, including your competitors.

That means more systems to integrate with, not fewer, and more surface area to keep current as those systems evolve. 

So the roadmap question isn't "Can AI help us build integrations in-house?" 

It's "Do we want our engineering team spending its AI-boosted velocity on integration maintenance, or on the product features that actually differentiate us in the market?" 

Every hour spent building, testing, certifying, and maintaining a connector is an hour your team isn't spending on the thing your customers actually pay for.

The bigger point

An AI roadmap is only as strong as the data foundation underneath it. 

You could use AI to help build an integration. You could use AI to help maintain it. And over time, AI will likely make that work faster. But Hapi can use that same AI.

The question is what you want to spend your AI-boosted engineering capacity on.

At Hapi, integration and data aren't a side project for our engineering team. They're our product. We've spent years building the normalized hospitality data layer, connecting systems at scale, developing vendor relationships, and handling the edge cases that don't show up in a prompt. And we've learned a lot of that through trial and error. We've made the mistakes, figured out why they happened, corrected them, and built those lessons into the way we work today.

That's experience your engineering team doesn't have to spend years acquiring on its own. The integration challenges, edge cases, certification hurdles, and maintenance issues that come with connecting hospitality systems aren't new problems for us. We've already encountered many of them, learned from them, and built around them.

That means your engineering team can spend its AI-boosted velocity on the things that make your product different… while we handle the connectivity and data foundation underneath it.

Tech companies may be using AI. So are we. The difference is what we're using it for.

Want the data model and vendor relationships handled for you, so your team's AI investment goes into product, not plumbing? Let's talk.

‍