There are no veteran artificial intelligence practitioners. A few companies and people might come to mind as pioneers, but for now it’s still very early days. Startups are pumping venture capital into these breakthrough technologies to compete against and even displace legacy companies. And those legacy companies are re-deploying their resources to stave off the startups.

The disruption feels familiar. In 2002, I left a major hotel chain to join Booking, which was just a tiny startup at the time. Ten of us held it together by a thread, with limited funding, but I took the risk because I believed the internet could fundamentally change how people experience the world—a transformation I had experienced myself when I moved from the Netherlands to the U.S. early in my career.

Today, Booking.com has more than 200 offices in 70 countries around the world--with more than 17,000 employees--and we find ourselves in another transformative moment, one that makes us feel like a startup again. The reason? The boom in AI and machine learning technologies. Like a startup, we’re experimenting with both--failing fast and evolving. That’s how you discover how these new tools can power memorable, perspective-changing experiences.

Image: BGC

Here’s what I’ve learned.

Get More Personal

Presenting inspiring ideas at the right time and at the optimal price is our goal, and advances in machine learning algorithms and AI help us do this in new ways. They leverage what we’ve gleaned about people’s preferences over the years while anticipating what they’ll want next—often before they even know they want it.

Today, machine learning touches every Booking customer, and this has been the case for many years already. Through our implementation and learnings, we offer increasingly more personalization, whether that’s on what type of property to stay in or how best to get to your hotel.

Build Connections

AI advances, such as Pinterest Lens, the company’s visual search engine, are connecting physical and digital experiences. Customers snap a photo of an object, anything from a lamp to a vegetable, and AI acts as an all-knowing personal shopper, finding the best match on Pinterest. Lens also curates items related to the first—tables and rugs to accompany the lamp, recipes for the vegetable.

It’s quite a technical feat. The search engine identifies clothing or furniture type, style, and other traits, a capability the AI is continually refining to expand its visual vocabulary. Lens then culls related items from billions of options on Pinterest.

This is the future of shopping, a seamless integration with AI that turns any discovery or curiosity in the physical world into a potential purchase.

Communicate Better

We live in a world where conversations about a brand and its products are ongoing, and AI is becoming critical, allowing companies to listen at a greater scale and with greater precision. A company can be vigilant as never before. The potential to respond to customers in real time is staggering.

The technology is also becoming adept at conversing. Today, our messenger bot, Booking Assistant, responds automatically to more than half of the customer inquiries it receives in English. It’s a welcome conversation; 80 percent of our customers prefer self-service to get the information they need.

Deploy Disruptive Tech to Optimize Performance

Machine learning and AI can also help head off bad experiences. Things go wrong. It’s inevitable and often unavoidable. But these technologies can optimize how we manage disruptions.

Qantas, one of Australia’s largest airlines, was one of the first to implement a data-science powered recommendation engine to instantly address and efficiently handle threats and interruptions to their operations. The system underwent a serious test in 2016 when heavy storms raged across Australia’s east coast. It performed incredibly well. Just 15 out of 436 Qantas flights were canceled, vastly outperforming competitors using the old manual system to manage disruption.

Rediscover Startup Roots—or Grow New Ones

Already, AI has taught us as a company important lessons about how to innovate. As we’re working to refine and build upon Booking Experiences, our push to help customers discover and book the best experiences throughout their trip, we are constantly testing with different goals and a specific mandate: match the speed of our digitally savvy customers.

We allow this team to test and explore possibilities (learning and failing fast). We are emphasizing that the goal is designing customer experience as much as it is about refining the technology—the two go hand in hand.

So, what will new companies do with AI? What will they try? My advice: Don’t wait for the perfect answer. As with much of today’s technology, AI will keep moving even if you and your business don’t. Like a traveler belatedly arriving at an empty airport gate, you don’t want to get left behind.

This article first appeared on LinkedIn.