IBM launches Bob, an AI coding platform with human checkpoints

IBM has globally launched Bob, an AI-powered software development platform designed to write and test code across the development cycle. The platform is already in use by more than 80,000 IBM employees after starting with 100 internal users in summer 2025. Bob introduces a structured layer that pauses for human-led checkpoints throughout the workflow. IBM says it has saved some teams up to 70 percent of time on selected tasks, equaling an average time savings of 10 hours per week. The platform supports multiple AI models including IBM's own Granite series, Anthropic's Claude, and some models from French AI firm Mistral. Unlike tools such as Cursor and Claude Code, which position the user at the beginning of a task, Bob pre-structures the development lifecycle into role-based stages where agents check in with users for approval at natural workflow checkpoints. Neal Sundaresan, general manager of Automation and AI at IBM, said in an interview that model capability alone is not enough. He emphasized that deployment, context structuring, and keeping humans in the loop determine whether AI actually delivers value. Bob is available in all regions where IBM does business. The pricing uses an internal credits system called Bobcoins, valued at one Bobcoin per $0.50 USD. Plans range from a 30-day free trial with 40 Bobcoins to an Ultra tier at $200 per month for 500 Bobcoins. An Enterprise plan is available through sales contact.