IBM Bob Agentic Development Platform Now Self-Hosted

IBM has announced the general availability of a self-hosted deployment option for its agentic software development platform, IBM Bob. This new offering enables enterprises to run Bob on their own infrastructure, including on-premises servers, private clouds, sovereign clouds, and air-gapped networks, ensuring that sensitive code and development context remain within the customer's controlled environment. IBM Bob is designed to cover the entire software development lifecycle, encompassing code understanding, work planning, execution of changes, and validation of results. The self-hosted deployment is available immediately to enterprise customers, who are required to source, license, and host a supported model themselves. IBM has not publicly disclosed pricing details, directing interested buyers to request a demonstration.

The generally available self-hosted deployment of Bob includes the core platform capabilities, such as its integrated development environment (IDE) experience, BobShell, parallel tool calling, the agent harness, and various skills and modes. It supports self-hosted, air-gapped, and hybrid model configurations. Customers can utilize a bring-your-own-license (BYOL) approach for eligible models from IBM's supported list. The platform also integrates with supported IDE and infrastructure environments. Additionally, optional Premium Packages are available for specific use cases like Java Modernization, IBM i, and IBM Z.

The architecture of Bob self-hosted separates the model layer from the platform, meaning that inference execution is determined by the chosen model's hosting environment. Customers select a model from IBM's supported list, which is categorized by deployment method. For self-hosted deployments, the customer manages the model on their own infrastructure. Supported models for this method include NVIDIA Nemotron and Poolside Laguna. For hybrid or private SaaS deployments, inference is handled by an external service via approved private connectivity patterns, with supported models including Claude Sonnet 5.0, Claude Opus 4.8, Gemini 3.7 Flash, and OpenAI GPT 5.6 Sol. This flexibility allows organizations, such as banks, to process core banking applications with on-premises inference while routing less restricted workloads to external models, all while developers interact with a consistent Bob experience.

This move by IBM addresses growing enterprise concerns around data security and regulatory compliance, particularly for organizations operating in highly regulated sectors or those with strict data sovereignty requirements. By allowing Bob to run in air-gapped environments, IBM is enabling advanced agentic development capabilities without necessitating the transfer of proprietary code or sensitive data to external cloud services. The platform's ability to integrate with existing enterprise IT infrastructure further facilitates adoption by minimizing disruption to established workflows and development practices. The dual approach of supporting both on-premises and hybrid model deployments provides a scalable and adaptable solution for diverse enterprise needs.