Webinar on Agentic Systems and Conversational Assistants

This webinar will present three real-world applications of generative AI. The webinar will feature three presentations: the first on designing agentic AI systems for production, and the next two on real-world applications of conversational assistants built with large language models.

• Davy De Waele – CTO at Ixor: Agentic AI has huge potential, but most AI projects never make it past the demo stage—something a recent MIT study makes painfully clear. This talk looks at why that happens and what it takes to design agentic architectures that hold up in the real world, with explicit attention to architecture, testing, clear ownership, change management, and day-to-day operations. Close collaboration with the customer remains key. Building agentic AI together as a system that can evolve over time, rather than handing over a black box and walking away, is essential to getting it into production and keeping it there.
• Mohamed Aziz Nebli – AI Engineer at Sirris: The project is about a conversational assistant built for nuclear engineers on-site who need accurate information quickly, without digging through technical files. Unstructured PDF documentation is structured into a semantic Neo4j-based knowledge graph that is also used as a vector store. One language model translates user questions into precise graph queries, while a second model turns the retrieved results into a clear, coherent answer. The system provides reliable, traceable responses through a simple cloud-based interface, optimized for real operational use.
• Laurent Christophe – AI Engineer at Sirris: The project focuses on developing a customer-facing conversational assistant for a company in the appliance repair sector. The assistant can answer general questions about the company, provide troubleshooting guidance for common appliance issues, and schedule repair appointments. The system integrates several components: translation to normalize multilingual user queries, semantic similarity ranking to detect user intent, structured LLM prompting to extract relevant information from unstructured user messages, and standard LLM prompting to generate natural, human-like responses.