Built on Feedback, Not Hype: Our Approach to AI at Open Destinations

Every product on the market seems to claim some kind of AI capability these days. It’s easy to add the badge — an assistant here, a copilot there. It’s much harder to make something that actually earns its place in someone’s day. So we sat down with our CEO, John Davies, and our Head of Engineering, Matt Watson, for a chat about how we’re thinking about AI in Travel Studio. What we’ve built, what’s coming, and why we’ve taken the approach we have.

The philosophy: assistants, not features

We started with the big question: what’s the point of all this?

“We believe the value is in building a series of digital assistants that make our customers and their teams far more efficient, effective, and productive,” John says.

Not one flashy AI feature bolted on for the sake of it, but rather a whole suite of AI-led assistants, each one quietly doing a different job across the booking flow, all working together in the same system.

Matt put it more bluntly: a lot of companies are adding AI just to be seen to have it, whether or not it does anything useful. “It’s not about whether we’ve got AI in the product,” he says. “It’s about whether what it gives back is genuinely useful and meaningful.” A good result, for him, is simple: someone spends less time on repetitive tasks and more time on the things that matter.

That’s really the thread running through everything we build.

It starts with the customer, not the technology

Ask either of them how we decide what to build next, and neither points to what’s technically exciting. They point to what customers are stuck on. Every decision runs through the same filter: what real pain point does this solve.

You can see that discipline in the order we’ve built things. Product Assistant, our first AI release, wasn’t picked because it was the most ambitious thing we could build. It was picked because it solved the single biggest, most repetitive burden our customers face: loading and maintaining thousands of accommodation products manually.

And that same logic keeps shaping what comes next. Customer conversations, prospect conversations, and what we’re hearing across the industry.

Product Assistant: earning trust before earning time savings

Building our first AI assistant wasn’t just about proving the use case; it was about proving the system could be trusted.

Matt is honest about the engineering challenge here: getting AI output to a standard people can rely on takes real work. You need the right context behind every instruction, the right question being asked, and then a thorough check on whether what comes back is good enough to use.

Product Assistant tells you the quality of the data it’s found and when it was last updated. “It’s really your call what you then do with that data as the user,” Matt says. It’s a small thing, but a deliberate one; instead of asking customers to blindly trust whatever the AI hands back, the system shows its working and leaves the final call to a person. It’s not there to replace judgement.

“You could have it do everything end to end but it’s still a machine. It doesn’t have the experience and the learning that the person driving the bus has. So we always have some form of human in the loop that says: this is what I think it should be, please confirm.”

The assistants are built to make people faster and more confident, not to take them out of the decision.

What’s next: turning insight into something bookable

If Product Assistant tackled the admin burden of getting products into the system, the next challenge is arguably the bigger prize.

“The holy grail here is itinerary building,” John says. “The ability to create an itinerary from a natural language search and make that bookable — to give customers and their agents a first draft that’s pretty close to what they’ll eventually sell.” It’s also, he admits, “probably the most difficult bit to solve”, which is exactly why it wasn’t first.

Here’s the thing though, generating something that looks like a solid itinerary isn’t really the hard part anymore. The harder problem is making it real; pulling from actual service details, contracted rates and pricing, and turning a plausible-looking suggestion into something that can confidently be booked.

That’s the gap John sees between Open Destinations and a lot of the AI-native competitors out there.

“A lot of the AI-led startups in this space are building a really intuitive interface, a great way to search for and structure an itinerary. What they don’t have is the depth of capability in the mid- and back-office: structuring, pricing, selling via different channels.”

Our advantage, he argues, is combining that operational depth with AI that makes it simple to use, rather than a slick front end with nothing solid behind it.

The advice they’d give tour operators watching this unfold

We asked what they’d say to travel businesses trying to figure out how to prepare for all of this.

John’s answer wasn’t really about AI at all; it was about the basics.

“Regardless of what AI capability we build, it relies on well-structured, good-quality data. The AI agents are only as good as the data they’re relying on, in the same way a human would be. It makes sense for organisations to get their data in order first.”

Where this is heading

For John, what’s most exciting isn’t a single feature; it’s who gets to use any of this in the first place. Smaller and mid-sized travel businesses are often put off by the sheer effort of adopting a new platform: getting data in order, setting up pricing, sales channels, integrations. AI tooling, he believes, will make it far more possible for those businesses to adopt this kind of technology and get the benefits that come with it, without the upfront effort that puts so many people off today.

For larger, multi-brand operator groups, trusting AI means it needs to align cleanly with their own tech and data standards and help them hit the innovation benchmarks they’re already working towards.

For Matt, it comes back to something simpler – not losing sight of the point of any of it. A good result means someone spends less time in front of a keyboard doing repetitive tasks, and more time on the things that matter to their business.

None of this starts with “what could AI do?” It starts with what our customers are stuck on, and AI is only the answer when it’s genuinely the right one.

This is a space we’re going to keep putting real time, people, and investment into because we’re not done until it’s right, genuinely adopted, and adding real value for tour operators everywhere, whatever their size and wherever they’re based.