Floqsta Technology Blog  

November 2022

Coming from a travel technology background, the Floqsta team recognized that our venture into the social travel space presented a compelling technology challenge and opportunity to innovate.

The Floqsta team is ideally built to tackle complex data and workflow interactions with technology. The online travel tech landscape is characterised by the need to integrate and orchestrate multitudes of disparate third-party supplier systems, intelligent contextualization, and personalization of offers to the consumer. In addition, high volume online transactions with UX that maximize online conversion is an ever-evolving space. Further to that, online payments, blockchain, customer data, and analytics. It is an environment that gets architects and CTOs excited.

In setting out to bring something new to the market that is not an OTA (Online Travel Agent) offering, we knew we were adding to these challenges. By adding the fundamental social matching aspect, we are overlaying on top of the above tech environment with some additional aspects.

Floqsta set out to break new ground by answering the following questions: How do we intelligently connect users with one another? How best to enable them to interact and message? How best to deliver the experience in a native app context? When do we present them with travel products and which ones should they be? How do we deal with the scalability when we arrive at millions of users? What intelligence can we bring to users for the travel planning?

The Product Design Challenge

Floqsta’s technologists and data scientists took a step back and first put ourselves in the place of users and the experiences that they would expect in such a platform. From a product design perspective, the Floqsta team feels that we are embarking on something new and fresh. Naturally, there has been inspiration from popular dating and social apps but also other apps that provide engaging experiences for users.The product design process started with ideation around the concept and the team iterated through wireframes and various levels of fidelity designs.

Choosing a Tech Stack

With ambitious timeframes for an initial pilot (a four-to-five-month development period), our tech team consciously chose to be as lean as possible. We expect that we will evolve our process over time as part of the learning process so moved quickly to establish lightweight engineering tools such as Shortcut for managing a Kanban board and Notion for sharing ideas. We have ended up using some of these tools beyond the development process. For example, Shortcut is also used to manage our pilot launch co-ordination and plans.

Floqsta’s technology choices include a React Native based app, microservices in the Platform backend, predominantly written in Javascript, a NoSQL database, and a Python-based AI engine (see below). We are leveraging AWS services as much as possible for the time being, including using Dynamo DB. As we move forward with the development, we will be constantly evaluating the toolset that works best for each stage of the venture.

Bring on the AI

A key differentiating factor of the platform is the intelligence used throughout the platform, including the way travel offers are contextualised to the user and intelligence in the Floqsta bot to assist uses with trip planning. Most significantly, AI and machine learning is at the heart of the platform in terms to drive what we call ‘Floq forming’. In fact, we started building the AI piece before we started with the app implementation.

We’ve worked with our Chief Data Scientist, Alan Walker, to build an NLP foundation, essentially unsupervised machine learning acting on 3rd party and platform data to determine proximity between words and descriptors to understand similarities between users based on their profile and other descriptions.

On top of this layer, we are working on user analysis algorithms which utilises machine learning to predict probability of compatibilities by processing user profiles, messages topics and semantic similarity and interaction history. Beyond this, there are intelligent floq forming algorithms and orchestration/display optimizations to ensure the resultant set of floqs are well tuned for the users.

The Opportunity Ahead

As we prepare for a pilot launch, we are mindful that the technology challenges that Floqsta brings are immense but at the same time, represents an incredible opportunity to push the boundaries of a high volume, interactive user platform driven by AI and machine learning. We look forward to technologists joining us on this journey as we move through various stages of growth!