The Domino Data RevX 2024 London event on 13 June focused on artificial intelligence (AI) as a key theme, with IT leaders highlighting training with real users and the separation of AI from IT infrastructure as best practices.
When implementing AI to enhance business processes, IT leaders have various options to choose from. They can either build or buy AI capabilities, utilize APIs for AI functionality, run AI on-premise, or opt for complete SaaS applications. Regardless of the approach taken, training based on internal data is essential.
During the event, Raj Mukherjee, head of data science and AI at Direct Line Group, shared how the company leverages generative AI (GenAI) for its GreenFlag car breakdown recovery service. He emphasized the importance of involving users in building accurate AI models, emphasizing the need for a strong data foundation for success.
Decoupling AI from IT infrastructure was also stressed by Sebastien Conort, chief data scientist at BNP Paribas Cardif. Conort recommended IT leaders ensure their AI systems are robust enough to support changes in AI models and infrastructure without requiring changes in the IT setup. He suggested using platforms like SciKit-learn and open source software to keep costs reasonable while developing AI products.