Intro

This is the first article in a series that explores our journey of integrating artificial intelligence (AI) at PayFit. The journey is a testament to the transformative power of technology when paired with comprehensive education and training. As we embarked on this path, it became clear that empowering our teams with the knowledge and tools to leverage AI was crucial. By focusing on specific use cases - such as automating meeting recaps, accelerating content creation and assisting people in finding and summarising relevant information to answer clients better and faster - we demonstrated the tangible benefits of AI and sparked enthusiasm among employees.

Beginning with clearly defined applications enabled us to vividly illustrate AI's power in streamlining operations and supercharging productivity. This approach not only provided immediate value but also fostered a culture of continuous learning and innovation. As our teams experienced firsthand the efficiencies gained through AI, they became more engaged and proactive in exploring new ways to integrate AI into their daily workflows. Our commitment to education and training ensured that every employee felt equipped to contribute to this technological evolution, paving the way for a more efficient and innovative future at PayFit.

So, what's driving us to document this story? As a B2B SaaS company in the "application stack", there's significant appetite and pressure to implement AI, but uncertainty about where or how to begin. We hope our experience in this series of articles can provide valuable guidance.

How it started

We aimed to free up creative energy for more strategic initiatives by automating repetitive tasks.

In November 2022, OpenAI launched ChatGPT, sparking curiosity and innovation among tech enthusiasts, including those at PayFit. By February 2023, ChatGPT Plus had rolled out in the US, further fueling discussions about its potential impact on our work.

Fast forward to April / May 2023, and we initiated our first company-wide actions:

• Expansion of an internal Slack channel dedicated to educating and sharing information and updates about AI
• Drafting and sharing guidelines with employees around the use of AI, encouraging people to try this technology whilst being clear about the limits in terms of confidentiality and privacy (for example, personally identifiable information)
• Starting to discuss in detail with our tools stack/partners their AI roadmap
• Discussing the first areas of investment for our roadmap

PayFit's deeper exploration of generative AI took off in the summer of 2023. We recognized AI's potential as a game-changer, understanding that beyond the impressive advancements in Large Language Models, the real impact lay in applying and integrating these technologies into our key routines and processes. This approach promised to deliver noticeable improvements quickly.

We decided to dive deeper into exploring key use cases where we thought AI could provide benefits, both internally and directly for our customers. 

We aimed to free up creative energy for more strategic initiatives by automating repetitive tasks and streamlining information retrieval. Our goal was clear: enhance operations, starting with a pilot program involving a subset of the Customer Success team in France, and then gradually unlock new AI use cases across the company.

Guiding principles

To achieve our goal, we established a set of guiding principles. We believe there's a middle ground between the ultra techno-optimists and the AI doomers, and it’s our shared responsibility to define it. It was therefore crucial to clearly articulate PayFit's AI philosophy and ensure it was integrated into our actions.

1. Companionship: Assisting and enhancing humans, not replacing them. We're optimistic that artificial intelligence can partner with human intelligence to improve how we work. This philosophy guides both our internal and external approaches.
2.  Experimentation and training are essential for grasping new concepts. We believe that continuous learning is key and the ongoing developments in AI technologies represent a significant need to educate and train existing and future employees. 
3. Expertise meets AI power: The importance of verticalization and use cases. We believe that beyond the remarkable capabilities of large language models, integrating them into company workflows through a variety of specific use cases will unlock incredible opportunities.
4. Security and safety is our priority.  We handle incredibly sensitive data, and when it comes to artificial intelligence, we are committed to extensive internal testing first. We will not compromise on critical aspects such as accuracy, privacy, safety, and security. We firmly believe in adhering to legal standards.
5. Experiment and measure. When implementing and using AI in workflows, it's crucial to measure its impact both qualitatively and quantitatively. This dual approach ensures a comprehensive understanding of how AI influences productivity gains and customer satisfaction (or other metrics). Our measurement protocols, similar to experimentation and A/B testing, involve forming hypotheses, developing test protocols, validating results, and leveraging feedback loops. 

How we see AI at PayFit

We were determined not to let Artificial Intelligence become just a buzzword justifying wasting money “just to try”

Leveraging AI for Strategic Success

At PayFit, integrating artificial intelligence into our strategy and approach is not just a technological shift but a holistic approach to achieving our goals. While AI holds immense potential, it's not about having 50% of our engineers dedicated solely to AI development. Realistically, we are not a foundational LLM company; we began with less than 3% of our team working on AI foundations, pipelines, and platforms. Instead, it's about weaving AI capabilities into the fabric of our daily operations, empowering our employees to utilize these tools seamlessly. Integrating AI enables us to enhance productivity, innovate continuously, and stay ahead in a competitive landscape.

Balancing Technology and Mass Adoption

The true power of AI lies not only in the sophistication of the technology but more importantly in its widespread adoption, leveraging focused use cases and the understanding of its benefits across various expertise areas. At PayFit, we’re focusing on mass adoption of AI tools while maintaining specialized teams that dive deep into key areas of internal investment. This approach ensures that while AI is accessible to all, providing tangible day-to-day benefits, we also have dedicated experts driving innovation and pushing the boundaries of what AI can achieve. This balance between broad adoption and specialized focus is crucial for maximizing the impact of AI on our business. Ultimately, like with all initiatives, it's essential to outline and measure a hypothesis. For the broadest use cases, we put significant effort into assessing the impacts both qualitatively and quantitatively.

Seamless Integration into Workflows

For AI to truly make a difference, it must be seamlessly integrated into both existing and new workflows. This fosters adoption and ensures that AI is not seen as an isolated tool but as an integral part of our operational processes. A small, focused team may work on developing commonalities and reusable AI components, but it's the broader team that adopts and integrates these solutions into their daily tasks. This collaborative approach links back to our core strategy of enabling PayFiters to leverage AI in their everyday activities, driving efficiency, and fostering a culture of continuous improvement and innovation.

Not a buzzword, a way to serve our transformation plan

We entered the realm of AI with the firm conviction that it would fundamentally change how we operate, yet we were determined not to let it become just another buzzword justifying wasting money “just to try”. Although not all projects can yet have a tangible ROI that shows up clearly in our P&L bottom line, we are focusing on learning how to assess the impact of each project and closely monitor it. That’s an ongoing process and we are still learning. Every new initiative is grounded in a robust business case.

Through this journey, we’ve learned two crucial things:

• The impact of AI in our company seems to revolve around three main areas: time gained, quality improvements, and access to unprecedented data and insights.
• AI is an opportunity to completely rethink existing processes. It's not just about making incremental refinements but about achieving substantial optimizations by transforming how we operate entirely.

This disciplined approach enables us to ensure that our AI initiatives are both meaningful and transformative, continually pushing the boundaries of what’s possible.

The Unseen Backbone of AI: Data, Documentation, and Data Governance

The quality of data is paramount; the better the data, the better the AI's performance. Producing and maintaining high-quality data is the underlying challenge in AI development. Artificial Intelligence acts like a light, revealing gaps in data management practices that might otherwise remain hidden. Inconsistent data, poor documentation, and weak data governance become glaring issues when organizations attempt to deploy AI at scale.

Stay tuned for more

In the coming days, we'll share a second article that will present the use cases we are currently exploring, as well as operational tips to ensure a successful implementation...