Interview with a CTO: Loïc Berthou in the spotlight

What happens when a CTO starts with zero legacy code, no red tape, and no 12-month IT queue? If you’re Loïc Berthou, CTO of Qlarifi, the world’s first Buy Now Pay Later credit bureau, you build a real-time data platform from scratch and still find time to debate whether using ChatGPT counts as cheating.

In the below interview, Loïc talks us through the realities of building tech without baggage; how his role has shifted from solo coder to strategic operator; why clean infrastructure beats clever hacks; and what every company should think about before throwing AI at their data.

Also, will he be replaced by AI? We asked. He answered. Read on for our full chat with our favourite Frenchman.

What does the role of CTO at Qlarifi look like?

My role has changed quite a bit since we started Qlarifi. I used to be the main software developer, but now I spend most of my time making sure our technology aligns with our business goals. I help our product team understand the technical side and create a roadmap that fits both our immediate and long-term plans. I also talk with clients, partners, and investors to represent Qlarifi's product and technical expertise.

When you code, is it the same as switching between French and English?

Not exactly. I think of coding more like playing with grown-up Lego. You start with an idea and put together different blocks to build the final product. It rarely turns out exactly as you first imagined, but that's part of the fun – figuring out the challenges along the way. It's incredibly satisfying to see the finished product come to life, get released, and help our customers.

When you start with 'a blank page' and a really tricky problem, how do you work through your thinking and start the development process?

I really enjoy tackling tough problems, especially with the smart people I work with. Usually, a big problem can be broken down into smaller, more manageable pieces. So, that's the first thing I'd recommend. Then, we brainstorm ideas on a whiteboard to discuss different approaches. It's important to be open to all suggestions – even the crazy ones can lead to the best solutions. Finally, we experiment. The first idea isn't usually the best, so quickly testing our ideas helps us learn and improve until we solve the problem.

Is it harder to start fresh like you did at Qlarifi or to work within existing infrastructure at large companies?

Like most things in engineering, it depends. Having a large existing infrastructure can be really helpful for launching new products, and it can feel less daunting than starting from scratch. However, I love a good challenge, and I believe that not having any existing infrastructure was an advantage for Qlarifi. It allowed us to design a cutting-edge credit bureau using the best technology available today. This has also enabled us to build an incredible product very quickly, which might have been impossible in a larger, more established organisation.

When you write code, how do you balance commercial ROI, privacy/security, and user experience?

At Qlarifi, security and privacy are always our top priorities and are central to what we offer, so they're directly linked to our return on investment. We believe that building these considerations into our product from the start actually makes things easier than treating them as an afterthought. It's also our mission to make the experience as smooth as possible for our users, both through our products and by providing excellent user experience, training, and support.

They [being The WSJ] say that software development is one of the roles most at risk of being replaced by AI, would you agree?

If you think software developers only write code, then absolutely! But the software developers I admire do much more than just coding. A more accurate headline (though less exciting) would be that software development will be significantly changed and enhanced by AI. I believe the day-to-day work of a typical software engineer will evolve in the coming years, with AI adding a new level of abstraction that will allow us to move from idea to product much faster. I see this as similar to when we moved from writing programs in 0s and 1s to using high-level languages like Python. This could actually lead to more "software builders" than ever, but perhaps fewer full-time expert coders.

If a complete novice decided to learn to code, what are your three top tips - and is using ChatGPT cheating?

1. Choose a project you're excited about building, like a tool you or your friends could actually use. Staying motivated is key, as programming can be frustrating when you encounter bugs. It's also really rewarding to see others use and enjoy what you've created.

2. Pick a language (like JavaScript or Python) and stick with it until you're comfortable. You might feel like you're missing out by not learning other languages, but don't worry about that. Focus on getting good at one language and building more complex features in your project to develop a strong understanding. Exploring other languages will be much more fun and rewarding later.

3. Find your community. Despite the stereotype, I think the best developers have a strong network that helps them grow. While I enjoy coding alone, I also love collaborating with other engineers to design new products, solve tricky bugs, or learn about exciting new technologies.

4. Using ChatGPT or other similar tools isn't cheating. They can help with prototyping, solving issues, and writing repetitive code. However, be careful not to rely on them too much, as they can still make mistakes. Make sure you understand their output and double-check the code before it goes live.

What is the number one thing a company can do to create value from its data?

Now more than ever, it's crucial for every company to realise that their data is significantly more valuable when it's well understood and organised. [Friday note - we didn’t pay him to write this!]

This was true before but this is becoming even more critical as companies want to leverage AI into their organisations. Without a very solid data governance, your AI strategy is going to be limited at best, wasteful at worst. On the topic of data quality, I'd recommend to read about the "Shift Left" approach and the concepts of Data Mesh & Data Products.

To read more about the exciting things Qlarifi is doing visit their website: https://www.qlarifi.com/