Journey Beyond Paves the AI Path to Boost Efficiency and Productivity

The Challenge

The rapid growth of generative AI (genAI) has transformed a wide array of business verticals, including the tourism sector. Taking a cue from this, the Executive General Manager of Technology of Australian experiential tourism group Journey Beyond, used AI capabilities to improve accuracy and efficiency organisation wide.

“Our exploration of AI involved recognising broader market trends, identifying opportunities, and leveraging it to address existing challenges. AI skills are becoming essential across our industry, whether you are a contact centre agent or part of the corporate team as mastering these tools greatly enhances efficiency and productivity,” says Madhumita Mazumdar, Executive General Manager of Technology at Journey Beyond, the Adelaide-headquartered company with 16 brands spanning tours, attractions, and unique accommodations.

Recognising the need for a balanced approach (protecting sensitive information while ensuring innovation and productivity), Mazumdar began exploring viable approaches and AI solutions in October 2023.

“Microsoft was rolling out new functionalities within its Azure ecosystem, which we predominantly use. Our strategy involved assessing these features to build a robust internal solution. After comprehensive discussions to outline our strategy and identify potential partners in the market, we collaborated with Agile Insights, a Microsoft partner, to develop a prototype,” she says.

Initially, Mazumdar’s focus was to create a prototype that operated solely within the company’s secure Azure environment, utilising standard chat activity without incorporating any external data. This ensured that all operations remained within the controlled environment, enhancing data security and reliability.

IBRS recommends enterprises dealing with highly confidential data must develop custom genAI solutions within a private data centre or secured cloud infrastructure. This approach provides an added layer of security and ensures that sensitive data remains under the enterprise’s control, adhering to relevant privacy laws and regulations.

It also advocates implementing secure encryption methods to protect data and limit access to sensitive information when interacting with generative AI chatbot solutions or other AI applications. Besides, employees must be educated about the risks of genAI and the potential for fake content or data breaches. Encourage them to report suspicious content and be vigilant in identifying and reporting potential threats.

“Following the initial development, we explored opportunities to enhance the system further by training it with our internal data. We began with our company policies, specifically those from Journey Beyond, to evaluate the outcomes. It’s important to note that all these developments are intended for internal use only, aimed at empowering employees rather than being customer-facing,” says Mazumdar.

The prototype developed in collaboration with the vendor proved to be of immense help as it allowed the company to quickly ramp up its team’s capabilities.

“By the time our team was trained, our partner was already making significant progress on our behalf. The insights we gained from the prototype were instrumental in enhancing our understanding of the technology. As a result of this experience, we now have team members who are more skilled in AI technologies, enabling us to take everything in-house,” Mazumdar says.

Aligning Generative AI with Business Needs

One of the persistent issues in the company’s contact centre was the overwhelming amount of data that agents needed to remember or access. By integrating AI solutions, the company aimed to streamline access to this data, ultimately enhancing operational effectiveness.

Mazumdar developed a chatbot, trained with data from Journey Beyond, and deployed it in Melbourne Skydeck’s (a part of the company’s portfolio of experiential tourism brands) contact centre. This tool enables customer service agents to quickly retrieve accurate information while interacting with customers.

“The implementation has proven to be a perfect solution for addressing our challenges. In June 2024, we introduced this technology to our contact centre leaders, and the feedback has been overwhelmingly positive. One of the key improvements we’ve noticed is in efficiency, as genAI enables us to communicate with customers in a more tailored manner, which allows us to deliver the best service possible,” Mazumdar, who wants to start benchmarking its impact in approximately six months, says.

The other bot has been developed to equip employees with the HR information they need at their fingertips. For example, when staff members have inquiries regarding parental leave or study assistance approval processes, they can now quickly retrieve the necessary details without having to consult a business partner. This streamlines the process, minimising the time spent searching for answers and reducing the likelihood of disseminating inaccurate information.

“It is truly remarkable to witness the overwhelmingly positive response from our HR department, as they have found the tool to be an invaluable asset in providing precise and relevant information to our staff,” says Mazumdar.

Refining the AI Solution

The journey from conceptualisation to implementation took Journey Beyond approximately five months. During this time, it focused on refining the user interface to ensure it met the desired aesthetic and functional standards.

Mazumdar says, “We initially rolled out the tool to a select group of users, including executives, general managers, and senior management. The goal was to ensure the tool was robust enough for senior leaders before expanding access. Building the underlying infrastructure was relatively quick, especially with the support of Microsoft documentation. If your team is familiar with Microsoft, getting started in that environment is straightforward”.

However, the more time-consuming aspect was ensuring that the tool provided accurate data.

“It was crucial that the tool did not fabricate information, such as generating a cancellation policy if it didn’t have the relevant data. Instead, it should clearly state when information is unavailable. For instance, one of the challenges faced was how the tool interpreted structured data, such as tables in our policy documents. Unlike humans, who can synthesise information from multiple tables, the tool struggled. We had to rewrite parts of the policy documents to ensure clarity and improve the tool’s understanding of the content,” says Mazumdar

She also encountered significant issues with a structured PDF document that included images and text. The tool inconsistently reads the text, leading to unsatisfactory outcomes. To address this, the document’s structure was improved to facilitate better data extraction and interpretation.

“The development process involved extensive work not only in building the tool but also in refining how it reads, indexes, and interprets data. Developers spent considerable time modifying the tool to enhance its contextual understanding. When users interact with the tool, context plays a crucial role; if a follow-up question is unrelated to the initial context, the tool may provide irrelevant answers,” Mazumdar says.

While working with our advisory clients, IBRS has noticed similar issues with document processing and data extraction. However, recent developments in document processing models have helped improve performance.  IBRS notes that with the integration of graph RAG, synthesising information across multiple tables becomes reliable.

IBRS observes AI will alter how HR activities are performed, from workforce planning to recruitment, onboarding, training, succession planning, and measuring employee engagement. The tasks HR professionals will do in five years will look vastly different from today, although the desired business outcomes will remain consistent.

IBRS recommends IT leaders must engage with the employee engagement manager or equivalent to discuss recent employee engagement results, insights, and questions that need further exploration. They must improve continuous collaboration using AI-infused tools that can analyse data and provide intelligent recommendations, facilitating decision-making and effective collaboration among team members.

Technology leaders and HR should review how the adoption of genAI and low-code tools will change existing software development team roles and plan new career trajectories for staff. CIOs should ensure a structured AI usage/governance framework is in place to guide the responsible and ethical use of AI within the organisation.