Let me take you behind the scenes of how Alphaguard started. It all began in March 2024 when Adnane and I met through a global VC program hosted by Antler. Antler runs a 3-month program twice a year in Paris (and in other locations worldwide), where they bring together potential co-founders to collaborate and create something new. By the end of the program, a few select startups receive funding, and Alphaguard was one of them.
Adnane and I quickly found common ground and started working together. We tested our ideas rigorously, conducting hundreds of interviews and refining our approach. Adnane is a unique blend—a data scientist and a certified fraud examiner—who recognized early on that the rise of Large Language Models (LLMs) could help make fraud investigations more efficient.
In May 2023, Adnane wrote an article where he tested the capabilities of ChatGPT in understanding complex fraud scenarios. He put GPT-3.5 and GPT-4 models through 18 typical Certified Fraud Examiner (CFE) questions, and the results were impressive:
This capability is particularly valuable in fraud detection, where massive amounts of data can overwhelm even the best investigators. Just as a radiologist's eye can tire after reviewing hundreds of scans, potentially missing critical details, fraud investigators can benefit from AI’s ability to spotlight high-risk areas, allowing them to focus where it matters most.
We chose to focus on public procurement, where the stakes are incredibly high. Fraud in this area not only leads to financial loss but also damages public trust. Our first design partners are Development Financial Institutions, which are trusted by governments and private funders to support projects that might otherwise struggle to secure financing due to their complex and risky nature.
Since March, we’ve been growing our team, adding two AI Engineers, and working closely with our design partners. We’re excited to showcase our progress at the end of this year and are committed to making public procurement more transparent and fair, one step at a time.
Alexandre
