IBM brings AI coding agent Bob inside the firewall for private deployment

Many Australian businesses want to tap into agentic AI to upgrade legacy systems, build software faster, and automate technical debt, but strict compliance and data governance rules often hold them back. For banks, health services, and government contractors, piping intellectual property and sensitive customer data into third-party public clouds simply is not an option.

IBM has moved to solve that bottleneck by launching a self-hosted deployment option for IBM Bob, its agentic software development and application modernisation platform.

AI assistance without exporting sensitive code

Unlike traditional autocomplete utilities, IBM Bob coordinates multiple specialised agents to handle planning, implementation, testing, and system modernisation across the software development lifecycle.

With this update, organisations can deploy the platform on customer-managed infrastructure, including private clouds, on-premises data centres running Red Hat OpenShift, and completely air-gapped environments. This ensures that source code, application architecture, and proprietary workflows remain inside the business's security boundary at all times.

Teams can choose between running supported models directly in-house or using a hybrid setup that routes non-sensitive tasks to external model providers, all while maintaining complete administrative policy controls and audit trails.

Why this matters for Australian organisations

Data sovereignty and supply chain risk are increasingly top-of-mind for Australian leadership teams, especially under heightened scrutiny from regulators and upcoming changes to domestic privacy and cybersecurity frameworks.

For companies operating in regulated sectors, this launch delivers a few key advantages:

• Strict data residency: Code and system blueprints never leave local servers or private tenant boundaries.
• Modernisation of legacy platforms: Teams managing complex, bespoke architectures can point AI agents at internal codebases without risking data leakage.
• Controlled governance: Engineering leaders maintain oversight over model access, sensitive data filtering, and automated pipeline approvals before changes reach production.

As AI tooling transitions from conversational chatbots to autonomous agents touching critical IT systems, the ability to keep those agents firmly behind company firewalls is set to become a foundational requirement for enterprise adoption.