AI That Never Leaves the Building: How Self-Hosted IBM Bob Unlocks Modernization for Your Most Sensitive Workloads

Every modernization leader in a regulated enterprise can name the systems that AI has never touched: the core banking platform, the claims engine, the mission planning application, the clinical records interface. 

Control has been the main barrier. Source code, application context and build artifacts cannot travel to an external model service when sovereignty rules, security policy or classification levels forbid it.

On 30 September 2026, IBM announced general availability of self-hosted IBM Bob, a deployment option for its agentic software development platform. 

Self-hosted IBM Bob brings AI-assisted development and modernization to on-premises, private-cloud, sovereign-cloud and air-gapped environments. It gives CISOs, CIOs and enterprise architects a credible path to modernizing systems that were previously off limits to AI.

What Self-Hosted IBM Bob Delivers

Bob helps teams understand, plan, execute and validate development and modernization work across the software lifecycle. The self-hosted option carries the core Bob experience into customer-managed infrastructure, including the IDE experience, BobShell, parallel tool calling, the agent harness, skills and modes. 

Bob’s backend runs on a customer-operated Red Hat OpenShift cluster, whether on premises or in your own cloud account. Long-lived systems carry years of business logic and hidden dependencies, so Bob’s ability to map those dependencies and assess how a proposed change affects surrounding applications matters as much as code generation. 

Optional Premium Packages extend this further. The Java Modernization package adds dedicated modernization workflows, the IBM i package adds platform-specific connectivity, tools and knowledge, and the IBM Z package adds enterprise-scale application analysis with IBM Z-specific context, all subject to licensing and deployment requirements.

You Decide Where Inference Happens

At general availability, you source and provide access to a supported model, and IBM groups the options by deployment method. Self-hosted models, which you install and manage on your own infrastructure, include NVIDIA Nemotron and Poolside Laguna. 

Hybrid or private SaaS configurations connect to an external model service through an approved private connectivity pattern and support Claude Sonnet 5.0, Claude Opus 4.8, Gemini 3.7 Flash and OpenAI GPT 5.6 Sol. A bring-your-own-license approach lets you build on eligible existing model investments.

Your model choice sets your data boundary. IBM states that with a supported self-hosted model configuration, code, development context and build artifacts can remain within the customer-managed environment. In a hybrid configuration, requests go to the external service you connect. 

A bank, for instance, can run on-premises inference for core banking development and connect an approved external service for a less restricted workload. We recommend classifying every application by sensitivity first, then assigning each one a model route that matches its risk.

What It Takes to Run It

Self-hosting shifts operational responsibility to your teams. You source, license and host the model, and you own OpenShift platform operations such as upgrades, scaling, availability, networking and identity.

IBM’s self-hosted documentation places security event logging and monitoring at the OpenShift platform level, so your SOC must wire Bob into its existing telemetry. A self-hosted label also does not certify an environment as compliant. 

Because the selected model and configuration determine where inference takes place, your security review should trace every data flow for each approved route, and your compliance team should validate those flows against the frameworks that govern your industry.

These requirements call for scarce skills: OpenShift platform engineers, GPU inference specialists, security architects and modernization leads. Organizations that hire IT talent for these roles face long searches, and many choose to recruit IT talent through a partner that already understands the stack.

ASB Resources supports clients with IT talent headhunting for these specialized profiles alongside our delivery teams, so your program keeps moving while the right people join.

How ASB Resources Gets You There

We combine Bob’s self-hosted capabilities with deep modernization expertise to deliver governed, secure outcomes for systems that AI could not touch before. Our engagements cover:

• Application estate assessment and workload classification by sensitivity and sovereignty requirements
• Model route design, from fully air-gapped inference to hybrid configurations
• OpenShift architecture, identity, access control and logging integration with your SOC
• Pilots on representative systems with measurable outcomes before wider rollout
• Roadmap alignment with IBM’s stated plans to expand the supported model portfolio and introduce multi-model routing

With a fully self-hosted model route, your code and development context stay inside your own environment, and we help you design the controls that let you demonstrate it to auditors.

Which of your mission-critical systems is still waiting outside your AI strategy?

Let the experts at ASB Resources help you identify the workloads ready for self-hosted IBM Bob and design the secure, governed architecture to modernize them. Schedule a call with one of our experts today!