IBM brings agentic AI to private infrastructure with self-hosted Bob

The shift addresses a growing tension between the desire for AI-powered development tools and the rigid security requirements of regulated industries. Many corporations manage mission-critical infrastructure where moving source code or data to public cloud services remains a non-starter. IBM’s latest move reverses the traditional AI workflow: instead of forcing organizations to move their data to the model, the platform now brings the AI to the data.

Neel Sundaresan, IBM’s General Manager of AI and Automation, emphasized that enterprise success hinges on sovereignty. By allowing companies to run supported models on-premises, IBM Bob provides a path for teams to leverage agentic workflows without compromising compliance. This approach aligns with broader market trends; research from the IBM Institute for Business Value indicates that 68% of executives struggle with data residency requirements, while hybrid and edge deployments are expected to claim nearly half of the AI infrastructure market by 2030.