Block Inc. has introduced Builderbot, a new internal AI orchestration tool designed to automate software development activities by coordinating multiple AI agents across the company’s codebase and interfacing with engineers through Slack.
According to Block, Builderbot carries out more than 200,000 operations each day and merges about 1,500 pull requests every week. The company says those merges represent roughly 15% of all production code changes across Block.
Engineers interact with Builderbot by tagging the bot in Slack and describing the work they need done. The system then proceeds to research, plan, and implement code for tasks such as bug fixes, service migrations, or new feature development, while team members continue to collaborate in real time.
Builderbot is presented as distinct from standard coding assistants because it is built to understand the full context of Block’s engineering environment. That includes awareness of every internal service, API, and company convention. Within that context, the system can pick up tickets from Linear and Jira, create branches, author code, open pull requests, monitor continuous integration pipelines, and iterate based on feedback.
Block emphasizes that Builderbot’s scope is limited to source code and system configuration. The tool does not access customer data, payment information, or personally identifiable information.
Brad Axen, Head of AI Capabilities at Block, described Builderbot as filling a gap between existing AI coding tools and large-scale engineering workflows: "The best way to think about Builderbot is as the missing layer between AI coding tools and how engineering actually works at scale," he said. "It handles the orchestration, the context, the environment, so our engineers can focus on the problems worth solving."
Technically, Builderbot is built on goose, an open source agent framework Block contributed to the Agentic AI Foundation under the Linux Foundation. Block also co-developed the Model Context Protocol with Anthropic, a protocol the company says has become an industry standard for linking AI agents to tools and data sources.
The rollout highlights an internal platform approach that ties AI-driven code generation to ticketing, branching, review, and CI processes within a controlled environment. Block describes Builderbot as an orchestration layer that executes tasks across multiple agents while preserving human collaboration and oversight.
Summary
Builderbot is Block’s AI orchestration tool working inside Slack to automate development tasks. It performs over 200,000 operations daily, merges about 1,500 pull requests per week and contributes roughly 15% of production code changes, operating only on source code and system configuration and integrating with Linear, Jira, and CI pipelines.
Key points
- Builderbot coordinates multiple AI agents across Block’s codebase and communicates through Slack, enabling engineers to request work by tagging the bot.
- The system executes more than 200,000 operations daily and merges roughly 1,500 pull requests each week, comprising about 15% of production code changes.
- Builderbot operates exclusively on source code and system configuration, and connects with Linear, Jira, and continuous integration workflows.
Risks and uncertainties
- Scope limitation - Builderbot operates only on source code and system configuration, so its capabilities do not extend to customer data, payment information, or PII; this limits what the tool can directly change.
- Coverage and adoption - The tool currently handles around 15% of production code changes, leaving the majority of changes outside its control and creating uncertainty about broader adoption across all engineering work.
- Integration dependencies - Builderbot’s function depends on connections with internal systems such as Linear, Jira, and CI pipelines, so any constraints or changes in those systems could affect its effectiveness.