IBM Brings Bob AI Assistant On-Premises for Secure Enterprise Use

IBM has expanded the deployment options for Bob, its AI assistant designed for application development tasks. The company announced last week that Bob is now available for on-premises use, a significant move for enterprises and government agencies that handle sensitive data. This development addresses a major limitation of previous versions, which required data to be sent to IBM's cloud for processing.

Bob, initially launched in April 2025, supports IBM i-specific languages like RPG, CL, and DDS, as well as SQL and COBOL. The latest version, Bob 2.0, introduced direct connections to IBM i-resident code and support for skills and modes, such as Db2 for i developer mode. However, its cloud-only operation remained a barrier for many potential users.

With the on-premises version, IBM now offers complete flexibility. Organizations can choose to deploy Bob entirely on their own infrastructure or use a hybrid model combining on-prem and cloud-based large language models (LLMs). The self-hosted option supports models like Nvidia Nemotron and Poolside Laguna, while the hybrid approach includes models from Anthropic, Google, and OpenAI.

IBM plans to expand the supported model portfolio and introduce multi-model routing in future releases. The on-premises version includes all features of Bob, such as the IDE experience, BobShell, parallel tool calling, and support for IBM i, IBM Z, and Java modernization. Interested users can learn more and download free trials at bob.ibm.com/.