Featured image courtesy of IBM.

IBM’s agentic software development platform, Bob, is now generally available for self-hosted deployment. Enterprises can run it on premises, within private or sovereign clouds, and in air-gapped networks. This move, first reported by MarkTechPost, gives companies with stringent data security and compliance requirements new capabilities. It also provides a clearer picture of how IBM is executing the “dual AI strategy” this publication observed in September.

Flexible model integration is central to Bob’s new offering. Clients can “bring their own model,” selecting between fully isolated options like NVIDIA Nemotron or Poolside Laguna for environments demanding complete network separation. Alternatively, hybrid setups allow integration with external models such as Claude, Gemini, or GPT. These setups balance security with access to leading commercial AI capabilities. This approach directly builds upon IBM’s recent enterprise AI efforts, including its partnership with OpenAI announced on August 14, 2026. That collaboration, alongside IBM’s focus on time series models on Confluent, led us to describe IBM’s strategy as dual, blending open and proprietary, internal and external solutions.

Making Bob available for air-gapped environments meets a need for sectors like defense, finance, and healthcare, where sensitive data cannot leave internal networks. These organizations often had to choose between advanced AI tools and security protocols. Bob’s self-hosted deployment removes this trade-off. It enables agentic software development within established security perimeters. This also extends to companies operating under data sovereignty laws, which mandate that data remain within national borders, often requiring private cloud or on-premises infrastructure.

Bob’s Reach Extends to Legacy Systems

Beyond its deployment flexibility, IBM targets its established enterprise customer base with optional Premium Packages for Bob. These packages extend the platform’s capabilities to Java, IBM i, and IBM Z modernization. This is a pragmatic play. Many large enterprises still rely heavily on legacy systems running on platforms like IBM i and IBM Z. Offering AI-powered development tools that integrate with and modernize these environments addresses a significant problem for CIOs managing complex, long-standing IT infrastructures. IBM delivers AI that works within clients’ existing realities, without forcing a complete overhaul.

The decision to support a range of models, from NVIDIA Nemotron and Poolside Laguna for strict isolation to Claude, Gemini, and GPT through hybrid setups, shows IBM understands the diverse AI adoption maturity levels and security postures across its client base. IBM is not betting on a single foundational model provider or deployment method. It offers choices instead. Enterprises can tailor their AI strategy to specific operational and security mandates. This contrasts with some AI providers who push a monolithic model architecture or cloud-only deployment.

Connecting the Pieces of IBM’s Enterprise AI

IBM’s August 14, 2026 partnership with OpenAI, which aimed to bolster IBM’s enterprise AI push, now appears as one facet of a broader, more nuanced strategy. While OpenAI models offer advanced capabilities, their cloud-native nature may not suit all enterprise requirements. Bob’s self-hosted option, with its support for air-gapped networks and diverse model integrations, is a complementary piece. It ensures that even the most security-conscious or regulated clients can access agentic development, either by using models designed for isolated environments or by connecting to external models under hybrid architectures.

This move solidifies IBM’s position as a provider focused on enterprise-grade AI solutions that prioritize control, security, and integration with existing IT estates. For many large organizations, AI adoption depends on practical deployment scenarios, data governance, and compatibility with established systems. The Premium Packages for Java, IBM i, and IBM Z modernization show this focus. They ensure Bob adds value where enterprises need it most, in the complex, often overlooked, world of legacy application management.

What is IBM Bob?

IBM Bob is an agentic software development platform designed to assist enterprises in building and modernizing applications using artificial intelligence.

What new deployment options are available for Bob?

IBM Bob is now generally available for self-hosted deployment, allowing enterprises to run it on premises, in private or sovereign clouds, and in air-gapped networks.

Which AI models can be used with Bob?

Enterprises can use NVIDIA Nemotron or Poolside Laguna for full isolation, or integrate Claude, Gemini, or GPT models through hybrid setups.

Compiled by Launch91 Desk from the sources linked above. More about Launch91.