AI at PayFit: Empowering Through Education and Innovation (Vol. 2)

Following our first article in which we lay out our guiding principles, this article goes deeper into the implementation part. Get here an overview of how most of our departments and teams initiated their transformation thanks to Artificial Intelligence. Latest articles will allow us to dig deeper in the technical aspects of key projects we conducted.

Exhaustive list of projects

AI is reshaping how we operate and it is our responsibility to prepare all teams to embrace this transformation

Our AI journey began almost one year ago. Since then, we’ve seen some highly impacting projects, promising in terms of how transformative they can be for our organization.

Today, we are closer than ever to our users thanks to artificial intelligence. 

  • Product managers and designers can now directly converse with our “AI-powered user personas” (LLMs trained to act like our personas). It works both ways: preparing for a user interview or determining which persona a user aligns with after a discussion.
  • Any source of business insights (Net Promoter Score, Customer Effort Score verbatims, customer tickets) can be exploited in ways that were not possible before (or required intensive human resources)

Our Customer Support department is currently empowered by artificial intelligence throughout their workflows.

  • Once a ticket arrives, it’s automatically classified and routed thanks to generative artificial intelligence. However, given the complexities of our business, it was challenging to successfully implement an automated solution (this will be detailed in a future article).
  • Agents are now assisted by AI to resolve tickets in the most efficient way possible.

Many other teams have found surprising ways to leverage this technology. Talent acquisition teams now use AI to combat bias in recruitment. Content teams use AI to enhance our Search Engine Optimisation (SEO) and content creation strategies. Meeting minutes are now automatically generated in a structured way.

We are convinced it is only the beginning and are excited to see what lies ahead of us. AI is reshaping how we operate and it is our responsibility to prepare all teams to embrace this transformation (Sales will be able to prepare their prospective calls like never before, users will see their experience on PayFit transformed by AI, and more!)

The importance of education

Unlocking AI's Potential for Non-Experts

By understanding the key principles of AI, business experts can unlock and apply AI to various use cases more effectively

Generative AI (Gen AI) presents an incredible opportunity for traditionally non-expert profiles such as business analysts, product managers, content experts, marketing professionals, and other domain specialists to harness the power of artificial intelligence. Traditionally, AI use has been limited to Data Scientists and machine learning (ML) Engineers who possess specialized technical knowledge. However, by understanding the key principles of AI, business experts can unlock and apply AI to various use cases more effectively. Their domain expertise enables them to identify valuable opportunities and tailor AI applications to meet specific business needs. This democratization of AI empowers more individuals within the organization to contribute to innovation and problem-solving.

Informing and Educating  All Levels

Successful integration of AI into a company requires informing and educating employees at all levels. Workshops play a critical role in this process, providing hands-on experience and practical knowledge about AI tools and methodologies. These educational initiatives help demystify AI, making it more accessible and less intimidating for non-technical staff.

Additionally, it is crucial to ensure that everyone is familiar with the roadmap of third-party tools used within the organization. This familiarity allows employees to leverage these tools effectively and stay aligned with the company's strategic direction. Sharing best practices and lessons learned from previous AI projects can further enhance collective knowledge and capability of the workforce. We’ve run several workshops with different teams, educating them about the basics of the technology, diving into specific use cases to connect it to their day-to-day and engaging them in conversation around AI. This doesn’t mean we’ve tackled every use case, but it established a foundation for understanding and prioritizing a broader range of potential applications. It's essential to do this at all levels of the company, from leveraging existing expertise to educating the entire leadership team on the power of these new technologies. 

Building a Community of AI Champions

Recognizing that different departments had different needs, Alex brought in key team members like Gaetan from Customer Success Operations and others from various functions, each bringing their deep expertise.

While hands-on support from the core AI team is beneficial, it’s not scalable in the long run. As the demand for AI solutions grows, it’s essential to develop a community of AI champions and decentralized project ownership. Encouraging proactivity among teams and fostering a sense of ownership over AI projects can lead to a more innovative and agile organization. By nurturing AI champions within different departments, companies can ensure that AI initiatives are driven by individuals who are deeply familiar with their respective fields, leading to more relevant and impactful AI applications. This approach not only spreads the workload but also accelerates the adoption of AI across the business.

Lastly, initiatives like AI Connect - a week of dedicated AI workshops led by internal and external experts - can foster collaboration and knowledge sharing across the organization. Held in early July 2024, AI Connect followed several training sessions and served as a centralized platform where employees could access resources, share insights, and seek advice from AI experts. By nurturing a collaborative environment, companies can ensure their AI initiatives are well-supported and best practices are effectively disseminated throughout the organization.

Ease of Implementation 

We needed to start with creating the basic AI pipeline or playground for developers and non-developers to test and play

One of the significant advantages of modern AI solutions is their ease of implementation. With the right infrastructure, accessing data and building the necessary pipelines and connections becomes straightforward. We've all experienced environments where we repeatedly stress the importance of data and documentation, yet it often proves challenging to engage the entire company in actions beyond high-level principles. 

We needed to start with creating the basic AI pipeline or playground for developers and non-developers to test and play. This ease of implementation enables faster experimentation and deployment of AI models, reducing time-to-value. 

By focusing on these key elements, companies can effectively harness the power of AI, democratize its use, foster a community of champions, and ensure that employees are well informed and educated about AI tools and practices. This holistic approach not only accelerates AI adoption but also maximizes its impact on business outcomes.

Insights on Implementing AI from our AI Operations Expert, Alex

As companies explore AI's potential, the path to successful integration isn't always clear. Alex, who has been instrumental in cultivating an AI-driven culture at PayFit, shares three practical lessons learned along the journey.

Deconstruct Before You Digitize

The first step in leveraging AI is not the technology itself. Rather, it's about meticulously breaking down what you’re trying to achieve. This analytical reasoning has nothing to do with artificial intelligence, yet the quality of the outcomes depends on it. Work with the business experts to identify your inputs, desired outputs, and the different steps to go from an idea to implementation.

Once this framework is established, implementing AI becomes a much more straightforward process. I often tell teams that prompting a solution is often less than 10% of the total time spent on building the solution.

Overcome Skepticism with Demonstration

I have often spent half a day creating an AI assistant for a team by only imagining the output they would need in its most basic form

It's common for teams to initially doubt AI's capabilities in their specific domain. You cannot blame them for that, this technology is so new and reshuffles so many preconceived ideas…

The solution?

  • First, you need to reassure them. This technology will greatly improve their life at work, give them access to insights they didn’t have before, help them gain crucial time, and bring clarity. However, it will not replace their expertise and creativity. 
  • Then, develop a proof of concept for them, even if it's not fully polished. When team members witness the potential firsthand, their skepticism often transforms into enthusiasm, paving the way for broader adoption. I have often spent half a day creating an AI assistant for a team by only imagining the output they would need in its most basic form. This approach usually triggers an "aha" moment, inspiring them to build something themselves.

Don’t rush decentralization

As previously mentioned in this article, having a few experts within the company mastering the technology is not scalable. The goal is to have each team autonomous and able to leverage this technology, but that doesn’t happen overnight! 

Generative AI is a highly intuitive technology, yet it appears complex at times and requires understanding a few key principles (prompting, context, architecture like RAG etc.) that are not necessarily easy.

You’ll need to provide active support and be on the lookout for people who can then act as your AI champions. By support, I mean being willing to explain the key concepts, taking the time to review prompts, challenging teams and explaining thoroughly what can be improved for the outputs a team expects.

Conclusion

The journey of integrating AI at PayFit has been a dynamic and transformative process, rooted in the principles of education, empowerment, and collaboration. By starting with specific, well-defined applications, we demonstrated the immediate benefits of AI, fostering a culture of continuous learning and innovation. This approach not only streamlined operations but also engaged employees across the organisation, inspiring them to explore and integrate AI into their daily workflows. Our commitment to educating every team member ensures that the entire workforce feels equipped to contribute to this technological evolution, laying the groundwork for a more efficient and innovative future.

Building a community of AI champions within PayFit has been pivotal to our success. By decentralizing project ownership and encouraging proactivity, we believe we have cultivated a sense of ownership and engagement among employees. This collaborative environment has allowed us to leverage the deep domain expertise of our team members, leading to more relevant and impactful AI applications. Initiatives like AI Connect have further facilitated knowledge sharing and collaboration, ensuring that best practices and insights are disseminated throughout the organization. This holistic approach has not only accelerated AI adoption but also maximized its impact on business outcomes.

As we continue to integrate AI into PayFit, we remain committed to our guiding principles of companionship, continuous learning, and responsible AI use. The ease of implementation of modern AI solutions has enabled us to experiment and deploy AI models quickly, reducing time-to-value and driving innovation across the company. By focusing on these key elements, we are confident that we can effectively harness the power of AI, democratizing its use and fostering a culture of innovation that will drive PayFit forward. The journey is ongoing, and we are excited to continue exploring new possibilities and unlocking the full potential of AI for our employees and customers.

We are not claiming we’ve done everything right, nor that all employees currently use AI every day, or that the measure of success is the number of AI assistants developed.  We are realistic; we are not yet an AI-first company, and our business is not about being a foundational model provider. However, we believe that magic happens when people combine the power of artificial intelligence with deep domain knowledge and specific workflows to achieve company objectives. We hope this perspective resonates with many of you.