How this travel company’s AI rollout improved 73% satisfaction: A 5-step playbook for your business

Follow ZDNET: Add us as a favorite resource on Google.

Highlights taken by ZDNET

• Agent AI tends to talk more than manufacturing services.
• Smart experts focus on use cases and supporting technologies.
• They test processes, refine methodology, and seek new opportunities.

Conversations with digital and business leaders about agent AI often revolve around the same sentiment: we’ve tested agents, but nothing is in production yet.

But while everyone is talking about AI testing, no business can run endless pilots without creating business value. And with experts suggesting experts who fail to exploit AI risk being left behind, there is a need to invest in effective agents sooner rather than later.

Also: How to build better AI agents for your business – without creating trust issues

At Booking.com travel expert Huy Dao, director of data and machine learning platform, is charged with delivering value from AI, including agent services. He has produced results by taking a systematic approach to service delivery, creating targeted solutions to the challenges clients face today and tomorrow.

Dao called this method in an interview with ZDNET as “connected travel,” where Booking.com tries to ensure that all aspects of the customer’s trip, whether it’s flights, hotels, or attractions, are considered as an integrated thing.

Also: Worried AI agents will replace you? 5 ways to turn anxiety into action at work

Creating a connected journey means working on different information. The data stack Dao’s team created allowed Booking.com to build new AI-powered services, including the company’s first agent, a partner-guest system that facilitates communication between customers and hotel partners.

Here’s what he’s learned so far, with five key lessons for other professionals looking to turn AI pilots into intelligent productivity tools.

1. Find a business challenge

Dao said the key to exploiting emerging technologies is finding the right use. While some experts remain uncertain about the potential of AI, he said that companies can use agent technology to overcome endless challenges.

“In my opinion, AI is not the flavor of the day, or even the year — it’s a real thing,” he said. “I see every day at work how AI can impact the way we do things.”

Also: 5 ways to use AI if your budget is tight

At Booking.com, Dao and his team identified that timely responses to customer inquiries was a major challenge for hotel partners. They realized that the agency’s technology could help hotels respond to inquiries faster and more efficiently.

“Before we rolled out agent solutions, whenever a customer wanted to connect with a hotel partner — for example, if you wanted to check if the hotel had a pool, or if you wanted to arrive an hour or two later — you would contact the partner and say, ‘Hi, can I have this information?'” he said.

“However, when the hotel staff answered, they often had to do more work to get the right answer. Also, sometimes they were not available when the customer asked the question. So, it could take several hours or more before the customer got the answer.”

2. Create a database

Dao said the data his team has created allows Booking.com to accelerate the adoption of AI and machine learning technologies in use cases, such as the one described above.

The Snowflake data platform is part of an integrated stack that includes ThinkSpot for analytics, Astronomer and Airflow for orchestration, Immuta for access control, Arize for machine learning visualization, and AWS for cloud computing. The data team also tests and implements AI models from major providers, such as OpenAI, Amazon Bedrock, and Google Gemini.

Also: Why enterprise AI agents can be the biggest insider risk

Booking.com’s best partner-guest communication system was developed in-house in Python, and the data team used LangGraph, an open source agency framework, to help the agent think about guest queries.

Dao said active agent systems are not limited to back-end systems. His team also thought hard about the user interface.

“We want to integrate technology or AI capabilities wherever it makes sense for our users,” he said.

“And in this use case, our partners already have a web-based portal to view their messages, so it was obvious that we had to integrate an agent right there to help them.”

3. Examine the use case carefully

With the business challenge identified and the technology platform optimized, Dao and his team focused on implementation, which occurred in two phases.

In the first phase, they developed a trusted assistant to help hotel partners deal with customer inquiries.

The result was an agent technology known as Smart Messenger, which collects partner, property, and reservation information to support hotel staff communicating with guests.

Also: 90% of AI projects fail – here are 3 ways to make sure yours doesn’t

At this early stage of the agent’s service, Dao said that the person is still very active.

“We want to make sure that the partner has the final say in how he wants to respond to customers,” he said.

“But we’re giving them an assistant, so instead of taking five minutes to respond, it might be a one-second click if they’re happy with what the agent is providing as a response.”

4. Refer others as confidence grows

Over time, Dao said confident hotel partners can start sending more work to the agent — and this phase represents the second phase of the agent’s implementation.

Here, Booking.com’s automated response tool allows hotel partners to define custom responses and create quick responses to guest questions, such as whether the hotel has on-site parking.

“This stage is when the agent says, ‘Okay, if you trust me enough, I can do it for you,'” Dao said.

“In this use case, the partner may be asleep when the customer asks the question, because it’s late at night. However, the agent can respond on behalf of the partner — and that approach helps in several ways.”

Also: 5 ways you can stop testing AI and start measuring it responsibly

Booking.com reported that early testing revealed a 73% increase in partner satisfaction compared to previous messaging tools. Dao said the agent continues to learn from past interactions and user feedback, adjusting its responses for accuracy and relevance.

“Now, with the agent, we measure the response to everything we do; we try it, and compare the improvement in satisfaction,” he said.

“Because the customer gets the answers they need, they don’t have to contact customer support, and that efficiency also lowers support costs.”

5. Look for other opportunities

Dao said that the exploitation of the agent must be combined with the crime of exploiting the individual. As his team sifts through customer experience, it continues to refine the platform, creating a foundation to support further agent testing.

“We didn’t want to build a stadium because of the platform,” he said. “When we built the platform, we thought about the user. We made sure we chose the right technology for the agent.”

Also: Is Google’s new $8 AI Plus plan worth it? How does it compare to the $20 Pro subscription

Dao said his team learned a lot from the agent development process. He also advised other experts to be aware of these studies.

“If you do your testing, you can assume that the agent’s system is good,” he said. “But when you go into production, things like latency can be a problem that you need to deal with. Then, you have to simplify your design and platform.”

Over the next 24 months, Dao expects continued growth at Booking.com. “You have to expect that, as a company, we will invest heavily in generative AI and agents, not just for the fun of it, but to enhance the user experience,” he said.

“People want an experience like ChatGPT now, and we want to have the same experience, or even better, when it comes to the experience of navigating our sites.”