Bringing agentic software development to sensitive and mission-critical workloads while giving enterprises control over their deployment, models and infrastructure.
Today, IBM announces the general availability of the self-hosted deployment option for IBM Bob, bringing agentic software development to sensitive and mission-critical workloads while giving enterprises control over their deployment, models and infrastructure.
AI is changing how enterprises build and modernize software but realizing its full value means applying it to sensitive code and business-critical systems. Security, data sovereignty and operational requirements may require source code, application context, and build artifacts to remain within approved environments. As a result, these constraints can leave critical applications outside AI-enabled workflows, slow enterprise-wide adoption, and potentially delay returns on AI investments.
IBM Bob is an AI-powered development partner that helps teams understand, plan, execute and validate software development and modernization work across the entire lifecycle. The self-hosted deployment option enables organizations to extend these capabilities to sensitive applications and workloads.
Developers can work through a consistent core Bob experience across supported deployment configurations, while platform, architecture and security teams align models and infrastructure with each workload’s security, sovereignty, performance and operational requirements.
At general availability, the self-hosted deployment can enable enterprises to:
Enterprise application estates include workloads with different security, sovereignty, infrastructure, and operational requirements. A single deployment architecture may not be appropriate for all of them.
IBM Bob’s self-hosted deployment brings its development and modernization capabilities into supported customer-managed environments, while giving organizations flexibility over where model inference takes place. They can run supported models on premises, including in air-gapped environments, or connect Bob to supported external model services through hybrid configurations. With a supported self-hosted model configuration, code, development context and build artifacts can remain within the customer-managed environment.
For example, a financial institution could use on-premises model inference for development work involving core banking applications while connecting Bob to an approved external model service for a less restricted workload. In both cases, developers work through a consistent core Bob experience, while platform and security teams align the underlying architecture with workload requirements. This flexibility can be especially valuable to organizations in financial services, government, healthcare and critical infrastructure, as well as enterprises protecting sensitive intellectual property, proprietary business logic and mission-critical software.
Organizations can connect IBM Bob to supported customer-provided models based on their deployment, performance and infrastructure requirements. At GA, customers source and provide access to a supported model for use with IBM Bob. The supported models are organized below by deployment method:
The selected model and deployment configuration determine where inference takes place and how code and development context are processed, enabling organizations to align the architecture with workload-specific requirements.
Organizations can also use eligible licenses for models on IBM’s supported model list through a bring-your-own-license (BYOL) approach. This enables customers to build on existing model and infrastructure investments while choosing the model and deployment configuration that best balances capability, performance and cost.
Looking ahead, IBM plans to expand the supported model portfolio and introduce multi-model routing in future releases. As organizations adopt more models, multi-model routing is intended to reduce the complexity of matching models to development tasks while helping teams balance capability, performance and cost.
Long-lived enterprise systems often contain years of business logic, dependencies and technical debt. Before modernizing them, developers need to understand how these systems work, identify dependencies, and assess how proposed changes could affect surrounding applications and services. Bob goes beyond code generation. It helps teams understand complex applications, plan multistep development work, execute changes, and validate outcomes. With self-hosted deployment, teams can apply these capabilities to supported applications that must remain within approved customer environments.
Organizations can extend this core experience with optional add-on IBM Bob Premium Packages for Java Modernization, IBM i and IBM Z, subject to applicable package licensing and deployment requirements. Each package addresses the distinct development and modernization needs of its respective platforms. IBM Bob Premium Package for Java Modernization adds dedicated modernization workflows. IBM Bob Premium Package for i adds platform-specific connectivity, workflows, tools and IBM i knowledge. IBM Bob Premium Package for Z adds capabilities for enterprise-scale application understanding and modernization, including IBM Z-specific context and application analysis.
Together, these packages provide a practical path for applying agentic development to new applications, and the foundational software the enterprise already depends on.
Bring agentic development to more of the sensitive and mission-critical software your enterprise depends on. IBM can help your organization evaluate supported deployment and model configurations, infrastructure requirements and optional Premium Packages based on your software application estate and development priorities.
Statements regarding IBM’s future direction, plans and intent are subject to change or withdrawal without notice, represent goals and objectives only, and are not a commitment, promise or legal obligation to deliver any material, code or functionality.