Block (Square, Cash App) just rolled out BuilderBot โ an AI agent that autonomously writes code, opens pull requests, and monitors CI across Block’s entire codebase. It’s already responsible for 15% of all production code changes. This isn’t a demo. It’s Harness Theory in production at enterprisescale.
An engineer tags @builderbot in Slack with a task description. The agent then autonomously:
1
Researches requirements and plans implementation
2
Writes code across the entire codebase โ an engineer on Cash App can modify Square services they’ve never touched
3
Creates branches, opens PRs, monitors CI
4
Iterates based on feedback โ picks up tickets from Linear and Jira
The key differentiator: BuilderBot understands Block’s complete technical context โ every service, API, and coding convention. It’s not a generic code assistant. It’s an agent with institutional knowledge of the company.
This is Harness Theory at scale. The human frames (Slack message with task description). The agent executes (research โ code โ PR โ CI โ iterate). 70/80 split โ exactly what Anthropic’s 400K-session data showed. Block just proved it works at 200K operations per day.
The MCP + Goose Connection
BuilderBot is built on Goose โ Block’s open-source AI agent framework, contributed to the Linux Foundation’s Agentic AI Foundation. Block also co-developed the Model Context Protocol (MCP) with Anthropic.
This matters because MCP is becoming the standard interface between agents and tools โ and Block is one of its architects. BuilderBot isn’t just using the harness layer. It’s building the infrastructure that the harness layer runs on.
Microsoft โ Copilot Cowork ($0.01/task, enterprise agent)
Pinterest โ Ask Pinterest (shopping agent)
Databricks โ Genie One (enterprise data agent)
Block โ BuilderBot (code agent, 15% of production code)
Seven agent launches in five days. The harness layer isn’t emerging. It’s arrived.
Business Engineer
My Life in the Harness + The Dynamo Doctrine
The seven harness principles. The primitive inversion. The 70/80 split between framing and execution โ now validated by Block’s 200K daily operations.
15% of all production code at a major fintech company is now written by an AI agent. 1,500 PRs merged per week. Tasks that took months take days. And it started with a Slack message. This is what the harness looks like when it’s not a framework โ it’s a shipping machine. The 70/80 split is real. The human frames. The agent ships.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur, HackerNoon and Search Engine People. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
![This Week In Business AI: The $195B Month [Week #9-2026]](https://i0.wp.com/fourweekmba.com/wp-content/plugins/contextual-related-posts/default.png?resize=150%2C150&ssl=1 "This Week In Business AI: The $195B Month [Week #9-2026]")
