The Magic in Mobi's Data





Isabella Roden


Image of rainbow mountains in Cusco, Peru

Mobi uses an exceptional collection of data—tens of millions of curated points of interest—to generate itineraries anchored in the realities of our constantly changing world, providing a magical experience for our partners and their customers.

If your customer wants to plan an incredible trip, they’re probably going to spend a lot of time doing research—an average of 20 to 30 hours, in fact.

Most likely, the average customer would use a search engine like Google to start exploring potential destinations. Then they’d have to decide on what activities they want to do, followed by what restaurants they want to visit, and which accommodations they want to stay in. They would need to confirm that any recommended hotels or restaurants are actually still open and then they’d need to check to see if they have any availability for their dates… and when the answer is no, they’d go back to the search engine again to look for other recommendations. From there they’d still have to decide how to best get from place to place, which is a whole process on its own. And they’d probably be left still wondering “did I make the right choices and did I find the hidden gems?”

Or maybe they’d become frustrated and decide to book their trip through a travel agent instead, who would then end up using the same lengthy and suboptimal research process to come up with a sample itinerary. If the customer decides to make any changes, say, leave on a different date, the process starts all over again.

In reality, being an expert about everywhere is really, really hard. In the travel industry, agents only get up to speed after months of training, and even then they usually only feel comfortable giving recommendations about the places with which they are truly familiar.

Whether it’s an individual traveler or a travel agent doing the research, all of these many hours of effort are focused on parsing through different forms of data about a trip destination—data on accommodations, routes, and experiences. There are millions of decisions to be made, and you don’t even know if the data you're getting is any good. Can you trust it?

It turns out that trip planning isn't just a planning problem, in many ways, it’s a data problem—you can't make a great plan without great data. Behind Mobi’s travel AI platform is a layer of extraordinary data that is so special that, well, there’s no other word for it but, magical.

Mobi’s AI platform is able to make amazing recommendations by pulling from an incredible data repository of tens of millions of points of interest (POIs). These POIs include information on things like places, accommodations, restaurants, experiences, and retail stores, but also unexpected and delightful data like the locations of wild animals, unusual rocks, and coral reefs. We collect this incredible assortment of data to help our partners’ travelers find and experience places where their best memories will be made and their hearts’ desires will be fulfilled. If that sounds like a tall ask, well, it is!

Image of Anna Jaffe

We've done the work of building a global content store of data so that our partners are able to deliver a magical experience for their customers that inspires travel during search and discovery—and then delivers on the promise when the customer is in-destination.

Anna Jaffe


Image of Anna Jaffe

We've done the work of building a global content store of data so that our partners are able to deliver a magical experience for their customers that inspires travel during search and discovery—and then delivers on the promise when the customer is in-destination.

Anna Jaffe


The amazing depth and breadth of our travel database is one of Mobi’s great strengths. It gives us the knowledge of a local combined with the computational problem solving power of AI. Part of Mobi’s magic is that we bring all of these types of information together quickly and easily so that any agent can act like a seasoned pro with detailed knowledge of any and all destinations.

It isn’t just that it’s magical for the customer—it also has to be magical for our partner. We make sure that it’s easy for our partners to integrate this database into their user interface (UI) and that the system is smartly tagged and that all of the data is legal to use. Perhaps our partner wants some types of POI data to not appear—they don’t want a CBD-infused breakfast place to appear as part of their family-friendly recommendations, for example. Our content store can allow for that as well. Feasibility is its own type of magic!

You may be wondering why we decided to collect all of this data ourselves instead of just using what’s already out there on the internet. Well, when we were first exploring how we could use our AI algorithms to make the most exceptional travel recommendations, we thought that we could use an existing data source from the internet to feed into our models, but we quickly found that there wasn’t any single source that was adequate to the task. For example, existing recommendations websites are often shaped around trying to sell certain experiences, which affects and limits what they recommend. This doesn’t serve the traveler or travel agent well. We’re trying to find the best results for what you’re looking for, not the experiences someone else most wants to sell you. So we decided that we had to make our own map of the world with access to unique datasets that don’t exist anywhere else in one place.

H3 cells showing POI (point of interest) distribution across a world map.

H3 cells showing POI (point of interest) distribution across a world map.

At Mobi, a POI isn’t just a GPS-coordinate. We expand and enrich a POI to include information about the buildings, the terrain, businesses, and how you might enter or exit the destination by different modes of transportation. For example, we download localized climate data to understand what time of day you might visit a certain location and use the climate conditions to shape our recommendation. We add information about seasonal changes to our knowledge of a location by overlaying hobbyist data on seasonal activities, like the timing of whale sightings for example. We can help you avoid rush-hour, interpret when hours of operation are strict or merely a “suggestion”, or time a long walk along a river for sunset. We can do this because we add these additional layers of data to our map of the world, which allows us to provide the context necessary to make well-timed plans.

Having this exceptional and massive collection of data allows Mobi to gain a deep understanding of a place. This means that for any specific location, we can know how that particular time of the day or year, weather, local events, or the interests of the user will shape our AI’s recommendation. At Mobi, we’re busy learning everything about every place at any time, to add that Mobi magic to our data and give you and your customers the best recommendations possible.

The result is an outstandingly planned trip anchored in the realities of our living, changing world—an itinerary full of incredible experiences and a schedule that has been shaped by that particular location and time. For a travel agent, this means they can quickly and efficiently provide an expert-level of service, and the traveler gets to experience the trip of their dreams.

AI Q&A: Anna Jaffe

As the CEO of Mobi, Anna Jaffe’s technical expertise as an MIT graduate and industry leader has shaped her vision to use technology to solve large-scale, intractable problems.

Read article

Image of night sky

It is a Big World

How to offer great experiences to all of your guests with an autonomous, real-time, planning solution

Read more about how we create plans