Block is putting a new AI message at the center of its engineering story. The company says its internal Builderbot system now helps manage code at a scale that reaches hundreds of millions of lines and hundreds of services across the organization.
That push carries extra weight because it arrives soon after Block cut more than 4,000 workers in a restructuring tied to its AI focus. The contrast has turned Builderbot into one of the clearest signs of how sharply the company is changing the way it builds software.
What Builderbot is doing inside Block
Block says 100 percent of its engineers use AI regularly in their day-to-day work, but existing coding assistants were not enough for the company’s full system. Tools such as Claude Code and Codex work well in a single repository, while Block has to coordinate work across a far larger software environment.
Builderbot was designed as an orchestration layer that coordinates multiple AI agents across the company’s codebase. According to Block, the system now handles more than 200,000 operations per day and merges around 1,500 pull requests per week.
That activity is equal to roughly 15 percent of all production code changes inside the company. Block also says work that once took months can now be completed in days.
The company says Builderbot can understand services, APIs, and engineering conventions used across its organization. That means an engineer working on Cash App can make changes to a Square service even without previous hands-on experience with that system.
Builderbot also pulls tasks from Linear and Jira, then carries them through the requested work. In Slack, employees can tag Builderbot with a short request for a bug fix or another task and follow the research, planning, and implementation in the same thread.
Why Block built its own tool
Block says the reason for building its own system was scale. The company explained that off-the-shelf assistants are effective within one repository, but they do not yet cover the full complexity of Block’s connected services and massive codebase.
Brad Axen, Block’s head of AI capabilities, described Builderbot as the missing layer between AI coding tools and large-scale software engineering. His comments make clear that the project is meant to connect technical workflows, not simply generate code.
Security remains part of the company’s framing. Block says Builderbot only operates on source code and system configuration, and does not access or process customer data, payment information, or personal identification data.
Part of a broader AI shift
Block says the last two years have been used to make AI a foundation for engineering work. That effort includes open-source Goose, joint work on the Model Context Protocol, or MCP, with Anthropic, and internal tools already used daily by engineers.
MCP is described as an industry standard for connecting AI agents to tools and data. In that context, Builderbot looks less like a standalone project and more like part of a broader technical strategy.
Jack Dorsey has said the launch is only the beginning of a longer series of AI tools Block plans to introduce. In a post on X, he said the company would begin talking more often about its “intelligence” tools.
That future-facing message sits beside a difficult staffing backdrop. Block’s latest restructuring followed earlier layoffs as well, including 931 job cuts in March 2025 and about 1,000 positions eliminated in January 2024.
The result is a company that is pairing deeper AI automation with a leaner workforce. For Block, Builderbot is not just another internal tool, but a signal of how its engineering model is being rebuilt around AI.
