Antoine Shagoury on AI in software development
From Implementing AI In The Real World — With Kyndryl's Antoine Shagoury · · Alex Kantrowitz
“We've deployed AI agent frameworks to improve the code lifecycle and deployment lifecycle for an international client in the Telco space, improving quality, lowering defects, and enabling faster, more effective releases.”
On , Antoine Shagoury, Chief Technology Officer at KYNDRYL HOLDINGS INC, spoke about AI in software development during Implementing AI In The Real World — With Kyndryl's Antoine Shagoury on Alex Kantrowitz.
About Antoine Shagoury
Antoine Shagoury, Chief Technology Officer at Kyndryl, discussed the company's approach to implementing artificial intelligence in a February 2025 interview. He described Kyndryl as having spun out from IBM over three years ago, bringing a heritage as implementers, integrators, and operators for managed services and infrastructure, and now partnering with Microsoft, AWS, Google, Dell, Nvidia, SAP, and Oracle. Shagoury stated that the company has deployed AI agent frameworks to improve the code lifecycle and deployment lifecycle for an international telecommunications client, improving quality and lowering defects. He described agents as "another step getting closer to really automated systems and automated processes." Shagoury identified data management as the biggest hurdle for companies working in AI, emphasizing the need for data inventory, lineage, management, and protection. He noted a material shift with more business leaders involved in AI decision-making, driven by a need for immediate solutions and less patience for long build cycles. He said there is less appetite for building homegrown AI solutions, advocating for a balance between off-the-shelf and open source models with micro models targeted to specific business needs. Shagoury characterized Kyndryl's partnership with Nvidia as one of the best engineering partnerships, focused on accelerating opportunity and shielding customers from complexity, and said that running workloads on Nvidia's environment yields exponential improvement in time, turnover velocity, and input accuracy.