Partner POV | Shifting from AI-assisted coding to AI-assisted delivery with IBM Bob

Article written and contributed by, IBM. 

AI has accelerated how code is generated, but bottlenecks were never just about writing code. They come from understanding complex systems, coordinating change across teams and managing risk across the software development lifecycle (SDLC).

As systems grow, development becomes less about individual tasks and more about coordinating change across interconnected systems. This shift requires a new model where work spans the full lifecycle instead of happening in isolated steps.

The need for a different approach

To address this shift, IBM built Bob in response to what enterprise teams consistently told us they needed.

Organizations are not just asking for gains in raw productivity. They need systems that reflect how outcomes are delivered across complex environments. Existing tools solve parts of this problem but still require teams to assemble their own systems across models and tooling.

Bob was designed differently: as an agentic SDLC partner that integrates orchestration, execution, and governance directly into the development process, helping teams move from isolated tools to coordinated delivery. Bob was built to optimize across quality, cost and performance while delivering trusted experiences for developers.

This approach is already being proven in practice. Within IBM, Bob has been adopted by over 80,000 users, with teams seeing an average of 45% productivity gains across complex, multi-step workflows.1

Enabling teams to work across enterprise systems

Bob is designed to adapt to how teams actually work. Built-in and custom modes allow developers to move seamlessly between planning, coding and reviews, while MCP integration connects Bob to the tools and systems teams already rely on.

As development becomes more distributed, maintaining context across systems remains one of the biggest challenges. By enabling reusable workflows, shared rules and deeper system awareness, Bob helps teams coordinate change more effectively without sacrificing flexibility.

Bob automates the mundane and augments the complicated. In practice, Bob can act like a junior developer for a senior architect to accelerate execution or like a senior architect guide for a junior developer to provide structure, confidence and clear direction.

Multi-model orchestration for enterprise AI

As AI adoption matures, organizations are navigating a growing set of tradeoffs around cost, performance and trust. The challenge is not just which model to use, but how to consistently get the best outcome across a rapidly evolving landscape. 
 
Instead of requiring teams to rely on trial and error to find the right blend of models and settings, Bob leverages a mix of frontier LLMs, open-source models, small language models (SLMs) and IBM's Granite SLM family to automate and augment the full software development lifecycle, from discovery and planning to design, coding and testing. 

Bob is built on the belief that beyond model capability, deployment architecture and user experience are critical to driving sustained, high-quality adoption. With pass-through pricing and visibility into usage and budgets, organizations can align spend with real outcomes.

Embedding security directly into development workflows

AI introduces new categories of risk that traditional controls were not designed to catch, from prompt injection to unintended data exposure.

To address this, Bob embeds security directly into development workflows. Prompt normalization helps prevent unsafe instructions, while sensitive data scanning and intelligent secrets detection surface risks as code is written. Policy enforcement is applied continuously, ensuring governance is maintained from development through delivery.

Modernizing complex systems with confidence

Modernizing enterprise systems has traditionally required piecing together changes across repositories, rewriting code incrementally and validating impact through trial and error. Even small updates can ripple across services in unpredictable ways.

Bob changes that experience. Bob helps teams build an understanding of system dependencies upfront, then executes coordinated changes across code, tests, and pipelines as part of a structured process. This allows teams to move from reactive updates to deliberate, system-wide modernization, even across legacy environments, such as Java, COBOL, PL/I and RPG. 

Client Zero: Scaling security testing in IBM

One example of improved efficiency comes from our internal revenue technology ("RevTech") go-to-market platform, a critical system supporting our global sales and client engagement across highly regulated markets. With complex architectures and strict performance requirements, testing and validation were both high-stakes and resource-constrained.

Bob was introduced to help teams improve performance and security testing while identifying more defects and vulnerabilities. This led to measurable gains in speed, scale and efficiency, demonstrating how AI can improve both the quality and reliability of enterprise-critical systems:

• 10x project-based ROI
• 300K payloads automated in testing scenarios
• Monitoring built in hours versus months

From internal success to real-world impact

These gains are not limited to our internal teams. Blue Pearl used Bob to accelerate delivery across its high-volume BlueApp platform. What typically required weeks of engineering effort was completed in just three days, as Bob streamlined analysis, refactoring and validation within existing workflows. This allowed teams to move faster while maintaining quality, demonstrating how AI can improve day-to-day delivery without additional overhead. This translated into measurable gains across delivery speed and engineering efficiency:

• ~90% faster delivery (30 days to ~3 days)
• 160+ hours saved through automated refactoring
• Zero defects post-deployment

"Working with IBM through Bob…enabled us to deliver measurable value." — Saireshan Govender, Group CEO, Blue Pearl

In other environments, the impact takes a different form. APIS IT used Bob to modernize mission-critical government systems spanning decades of technical debt, including mainframe and .NET environments. With limited documentation and complex dependencies, Bob enabled rapid system understanding, automated documentation and coordinated refactoring. The results show a step-change in modernization efficiency: 

• 10x faster architecture analysis and documentation
• 100% accuracy in documenting legacy JCL/PL/I systems
• Hours vs weeks to refactor legacy services into modern APIs

Maximizing the IBM portfolio

Bob can tap into the strength of the IBM portfolio to deliver better outcomes and increased sophistication. We are already seeing impactful ways IBM and customer teams are using Bob to maximize the value of IBM solutions, from optimizing agents in watsonx Orchestrate to modernizing applications on IBM Z and IBM i platforms. Later this year, we will introduce Premium Packages to extend Bob with platform-specific capabilities, including prebuilt workflows, domain expertise, and seamless integrations across IBM platforms.

Bringing it together across the lifecycle

The next phase of software delivery will be defined by AI-orchestrated systems, not isolated acceleration. Bob was built to support that shift, working alongside teams to connect planning, execution and validation across the development lifecycle.