IBM adds multi-agent controls and modernization workflows to Bob developer platform

IBM has updated IBM Bob, its agentic software development platform, with features aimed at making AI-assisted engineering easier to manage across large enterprise codebases. The July 9 release adds multi-agent capabilities, cost and usage analytics, and packaged modernization workflows for teams working on IBM Z, IBM i and Java applications.

The announcement positions Bob less as a code-completion assistant and more as an orchestration layer for software delivery work. IBM says the platform can match models to tasks, coordinate execution across agents, and expose productivity, quality, performance and cost signals through a new analytics feature called Bobalytics. That matters because enterprise buyers are increasingly asking not only whether AI can produce code, but whether teams can review it, govern it and keep spending predictable.

What changed

• Bobalytics gives teams a way to monitor AI consumption, allocate resources and keep oversight as AI development scales.
• Parallel, model-native tool calling lets Bob request and run multiple tools in a single turn, which IBM says can reduce waiting in multi-step engineering tasks.
• Subagents handle exploratory work in separate contexts, returning summaries to the main agent while helping control context-window growth and cost.
• Premium packages add repeatable workflows for IBM Z, IBM i and Java modernization projects.

The premium packages are the most enterprise-specific part of the update. IBM says the IBM Z package supports COBOL and PL/I modernization and JCL analysis, while the IBM i package adds remote file-system integration and workflows built around IBM i operational patterns. The Java package focuses on guided modernization, including migration to Java 25, refactoring and dependency analysis.

IBM also ties the release to a broader shift in the developer-tool market. As AI systems generate more code, organizations are finding that the hard work often moves downstream into validation, review, security checks and large-scale modernization planning. IBM cites a 2026 GitLab AI accountability report in saying that 85% of surveyed DevSecOps professionals agree AI has shifted bottlenecks from writing code to reviewing and validating it.

The update is available now, according to IBM, with Bob offered for download from the Bob site. For CyberOGZ readers, the key signal is that major AI coding products are moving beyond single-agent chat interfaces toward governed, repeatable workflows that can be measured and audited inside existing enterprise software environments.

Sources

• newsroom.ibm.com
• bob.ibm.com

Cover photo by Nemuel Sereti on Pexels, used under the Pexels License.