From prototype to product, Back Market has developed its own internal GenAI platform, Monster, aimed at being modular and secure.

The experience begins with the need to provide ChatBot-type experiences using AI. Born as a side project, the platform hosted on GCP and based on chainlit, evolves through experiments and business scenarios that arise. Monster is designed from the start to be modular, flexible, and independent of the providers used. It can connect to Vertex AI, OpenAI, or others, allowing the business to adapt Monster AI to needs effortlessly. However, Monster AI does not neglect the security and confidentiality of data, which are essential pillars of our implementation choices. We will show the challenges encountered and security features developed to ensure the scalability and reliability of AI projects for our 700 internal employees or by enhancing the productivity of more than 150 of our support agents. Despite the technical appearances, this has also been a human challenge, on which we will share our successes but also our failures, from responding to strong business expectations to neglecting a segment of users. Our session explores the challenges, breakthroughs, and implementation choices, from prototype to product, with the necessary hindsight to share our lessons and advice. Join us for an in-depth exploration of the challenges faced, significant advancements made, and valuable lessons learned.

Recorded in April 2024 at Paris, Palais des Congrès, Porte Maillot.