Steven Lemmens

Sarrai is an AI-driven platform for customer service, specifically built for European SMEs and scale-ups that want to relieve their support team without compromising on quality.

As founder, I am responsible for the entire technical product. Specifically, I build, among other things:

Underneath this lies a complete backend, API layer, data model, and integrations — which I design, build, and maintain myself.

For me, Sarrai is also proof of my approach as a freelancer: building end-to-end software, making architectural choices that will still be valid a year from now, and linking technical decisions to business impact. No theoretical advice — I build it myself, every day.

Technologies used:

I was responsible for the complete support of customers once they had completed their onboarding for the ERP system. The team soon grew, which allowed me to manage colleagues who were also in the support department, and I also had the opportunity to think more about Customer Success and customer satisfaction itself. Thanks to the integration of AI, we were able to quickly handle 30% of customer inquiries fully automatically with a chatbot (to the great satisfaction of the customers), and more focus was placed on AI.

After a year of intensive focus on the support process and also taking over the documentation of the entire ERP platform, we were able to increase the use of AI, and it was capable of independently handling up to 50% of all customer inquiries, resulting in high customer satisfaction.

Here I combined technical expertise with a strong understanding of customer needs and business processes.

At Optedo, I started as a software engineer to work on their cloud ERP platform in .NET and Angular. Here I worked on the core features of the software package, such as time registration and invoicing, and I developed integrations with, among others:

Technologies used:

For an application that had hundreds of users daily, performance was an absolute must, which led me to specialize in debugging queries and optimizing performance bottlenecks.