More than 80,000 IBM employees are reportedly already using Bob. The AI assistant, which we should really call a coding “partner” according to IBM, is said to offer a 45 percent boost in productivity. Now it’s up to the company’s customers to see if that efficiency is in store for them as well.
After months of internal use, IBM Bob officially entered general availability today. The road to this point was not without its bumps: Bob 1.0.0 was released on March 24, but security researchers had already revealed in January that the platform could be manipulated via the CLI to execute malware. The IDE also proved vulnerable to common AI-specific data leakage vectors. After that false start, today is Bob’s chance to prove itself once again.
Bob combines leading LLMs, open-source models, and IBM’s own Granite SLM family to support the entire software development lifecycle, from planning and design to coding and testing. The platform automatically selects the most appropriate model for each task. IBM aims to eliminate choice overload for developers, though the “black box” approach also draws skepticism from developers who want to know what’s happening behind the scenes. According to IBM, security is built into the workflows, intercepting risks such as prompt injection. Palo Alto Networks previously offered integration with Prisma AIRS for additional security scanning of prompts and code within Bob. It is quite possible that organizations and system integrators will want to add more security tools to Bob and similar AI assistants for their own use cases.
Mainframe module and pricing
Concurrent with the GA launch, IBM is introducing the Bob Premium Package for Z. This package integrates with IBM watsonx Code Assistant for Z and offers an Architect mode for analyzing application structures and dependencies, plus a Code mode that can generate and refactor COBOL, PL/I, and Assembler code. For now, customers will receive it as a free, private technical preview. The package targets mainframe environments struggling with technical debt and knowledge aging. IBM also announced that on-premises deployment will follow later this year for organizations with strict data residency requirements.
For the regular version, IBM uses a so-called “Bobcoin” system, which is essentially credits. The Pro tier costs $20 per month and provides 40 Bobcoins; Ultra costs $200 and includes 500 Bobcoins. One Bobcoin is roughly worth fifty cents, but becomes more cost-effective at larger scales. The pricing model resembles that of other usage-based systems in many ways, such as GitHub Copilot, which will be launched in a few months.