- | 6:00 pm
ENTERTAINER introduces HAPI, an AI-powered personalisation assistant
The 25-year-old lifestyle app has introduced HAPI, a multilingual AI assistant built on AWS.
Dubai-based ENTERTAINER has launched an AI-powered lifestyle assistant, HAPI, developed in collaboration with Amazon Web Services, as the company looks to expand beyond its long-standing Buy One Get One Free model.
Integrated into the ENTERTAINER app, HAPI, short for Helpful Artificial Personalisation Intelligence, uses generative AI to help users search offers, plan activities, and handle routine customer service queries through conversational prompts. The assistant currently operates in Arabic and English and can respond in multiple languages commonly used across the GCC.
The launch marks a shift in how users interact with the platform. Instead of manually browsing or filtering offers, users can ask open-ended questions and receive contextual responses, reflecting a move toward intent-based discovery rather than keyword search.
HAPI is structured around four specialised AI agents. A knowledge agent responds to frequently asked questions related to memberships, offer rules, and platform usage. A search agent surfaces relevant offers based on conversational inputs such as location or preferences. A planner agent combines multiple requests to generate full-day or weekend itineraries. More complex or sensitive queries are passed to human customer service teams through an escalation agent.
Founder and CEO Donna Benton said the assistant is designed to address recurring customer questions around pricing, usage rules, and participating outlets.
“There’s still a journey to go, and with any bot, it needs time to learn,” she said. “In the beginning, we put in as much information as possible, including the rules of use and outlet data, because those are the questions customers ask most often.”
Benton described the change as a shift in how users engage with the app. “Think of us as the new ChatGPT and everyone else as the old Google,” she said, referring to the move from keyword-based search to conversational interaction.
A FOUR-WHEEL AGENT WORKS
HAPI is built around four specialised AI agents. A knowledge agent answers frequently asked questions related to memberships and platform usage. A search agent retrieves relevant offers based on conversational inputs, such as location or preferences. A planner agent combines multiple requests to generate full-day or weekend itineraries. More complex or sensitive issues are transferred to human customer service teams through an escalation agent.
The knowledge agent serves as a reference point for the app, answering frequently asked questions about memberships, offer rules, and general platform information.
The search agent focuses on discovery, surfacing relevant offers and merchants based on conversational prompts such as location, preferences, or group type.
The planner agent combines multiple requests into structured itineraries, allowing users to plan an entire day or weekend, from dining and activities to experiences, based on budget and interests. When a request requires empathy or complex problem-solving, the escalation agent transfers the conversation to a human customer service representative.
Following a phased rollout late last year, HAPI has already generated more than 220,000 conversational exchanges. According to Benton, 50% of those interactions have been resolved without human intervention. “We’re probably about 75% of the way there,” she said. “But it’s only been two months. The proof is in the data…it’s going really well.”
Multilingual capability has emerged as a key differentiator, particularly in the GCC. The assistant can automatically detect and respond in multiple languages, addressing a long-standing limitation of traditional customer service models.
“Another nice addition to what we’ve had is all the languages, the multilingual that we can put in, because obviously not all of our customer service agents speak all the different languages,” Benton said. She added: “But when you can type it in, it’s a lot different.”
CHALLENGES BEHIND THE SCENE
Both companies acknowledged that launching HAPI required balancing speed, cost, and accuracy.
“We all like to do things faster, but it has to be good from an entertainer’s point of view. Obviously, there’s a lot more cost involved in something like HAPI. But for us, we want to give back to the customers. We want to give back to the merchants,” Benton said. She added, “It’s an investment and time. We always want to do things faster, and the information has to be correct.”
From AWS’s perspective, the challenge involved integrating generative AI into a platform with decades of legacy systems. “There is quite a bit of history and legacy in terms of systems, capabilities, services that have been delivered,” Chris Erasmus, General Manager for AWS in the UAE and wider Middle East and North Africa, said. He added: “Part of this process, which we have to be able to do, is marry that to what we’re trying to drive in terms of modernization as well, so quite often our teams would have to figure out exactly how we drive this modernization piece in order to get the outcome. That’s not an easy thing.”
The ENTERTAINER views HAPI as a foundation rather than a finished product. Insights from user interactions are expected to drive deeper personalisation for consumers and more targeted exposure for merchant partners, as the company looks to embed itself more closely into users’ daily decision-making.
“This is an intuitive service, so over time, as more consumers continue to adopt it, that will bring us more insights in terms of how we evolve the service as well. It is something that will continue to move pretty quickly in terms of the service offering,” Erasmus said.
