Block’s Builderbot now handles 15% of production code changes, automating routine engineering work across the company
Jack Dorsey’s Block has rolled out Builderbot, an AI-native engineering tool that the company says now handles roughly 15% of all production code changes. According to Block, the system runs more than 200,000 operations per day and merges about 1,500 pull requests each week — a sign of how AI is moving from suggestion engines into actual production workflows.
What Builderbot does
- Builderbot is not a single-repo coding assistant. It functions as an orchestration layer that coordinates multiple AI agents across Block’s entire codebase, working with services, APIs, internal rules and system patterns.
- Engineers invoke it inside Slack: they tag Builderbot, describe a task, and the system can create a branch, write code, open a pull request, monitor continuous integration, and respond to feedback.
- The tool only touches source code and system configuration; Block says it does not access customer, payment or personal data.
Scale and impact
- Per Block: ~200,000 operations per day and ~1,500 merged pull requests per week.
- Builderbot now accounts for about 15% of production code changes.
- Block reports that 100% of its engineers use AI regularly, and Builderbot enables teams to make changes even in parts of the system they haven’t worked on before.
- The company says tasks that once took months can now take days, with Builderbot handling scaffolding and repetitive work while engineers focus on product decisions and direction.
Why Block built it
Brad Axen, Block’s head of AI capabilities, described Builderbot as “the missing layer between AI coding tools and how engineering actually works at scale.” The rollout follows broader restructuring at Block, where Dorsey previously said the company was cutting headcount and reorienting around AI and smaller teams — a shift that included more internal reliance on intelligence tools.
Industry context
Builderbot highlights a wider trend of embedding AI deeper into fintech and crypto companies’ operations. For example, Coinbase has introduced an AI advisor that lets users manage portfolios via natural language and can connect to trading rules. But Builderbot is notable for operating inside engineering teams rather than being only user-facing.
Open questions
As AI agents touch more production code, companies face pressing questions about code quality, review standards and security. Block says humans still guide the process and make the product-level decisions; Builderbot is intended to handle lower-level work that slows development. How teams maintain rigorous review and security controls as AI takes on more responsibility will be a key test for this approach.
Read more AI-generated news on: undefined/news
