Block said agentic AI helped write and review nearly all of its production code changes in June 2026, while its Moneybot financial assistant surpassed 1 million weekly engaged accounts and Cash App Consumer Lending originations increased 59% to $18.9 billion, illustrating how the company is applying its artificial intelligence infrastructure to both internal operations and customer-facing financial products.
Block built its goose AI framework in early 2024 to work across different underlying models rather than relying on one AI provider. That architecture led to Builderbot, which orchestrates AI activity across Block’s codebase and helped support the high level of AI involvement in production coding during June.
The company has extended the same approach into additional internal systems. Managerbot is already automating tasks including marketing, margin analysis and operational fixes for sellers, while Block launched Buzz in July as an internally developed system for agent collaboration, communications and code repositories.
Those applications operate through one internal platform for model access, routing, tools and permissions. Block said its evaluation systems allow it to improve the quality of its AI applications independently of the underlying models and switch to better models as they become available.
Moneybot represents a customer-facing expression of the strategy. The financial assistant is generally available and has surpassed 1 million weekly engaged accounts, which Block defines as accounts that sent at least one message to Moneybot during the trailing seven-day period.
The AI investment is being paired with rapidly expanding financial-services activity across Cash App. Consumer Lending origination volume increased 59% year-over-year to $18.9 billion, driven primarily by Cash App Borrow, while Block said risk loss rates remained healthy.
Block defines Consumer Lending originations as lending through Cash App Borrow and its buy now, pay later products. The company said origination volume has more than doubled over the past two years while maintaining healthy risk-loss rates.
Cash App Borrow is also being combined with other lending products. Block continued scaling Afterpay Post-Purchase and launched Afterpay Pre-Purchase to general availability, allowing customers to choose dynamically between debit and BNPL before eligible transactions. By June, BNPL represented 17% of card spending enabled for Afterpay Pre-Purchase, with adoption particularly strong in groceries, gas and utilities.
Block’s proprietary credit infrastructure also powers Cash App Score, a near-real-time individualized credit score based on Cash App activity. During Q2, the company expanded engagement testing for Cash App Score to additional customers.
The lending expansion has a corresponding cost. Transaction, loan and consumer receivable losses increased 99% year-over-year on a GAAP basis, primarily due to higher loan volumes, although Block said cohort-level Borrow risk-loss rates remained healthy.
Broader Cash App engagement also increased. Primary Banking Actives grew 17%, Cash App had 59 million monthly transacting actives in June, and Cash App Commerce Enablement volume increased 17% to $56.5 billion.
Block generated Q2 gross profit of approximately $3.17 billion, up 25%, including $1.97 billion from Cash App, up 31%, and $1.16 billion from Square, up 13%. Operating income reached $447 million.
Adjusted Operating Income reached a record $864 million, representing a 27% margin, while adjusted EBITDA reached a record $1.2 billion and adjusted diluted EPS increased 65% to $1.02.
The company raised its full-year forecast and now expects 2026 gross profit of $12.51 billion, up 21%, and Adjusted Operating Income of $3.47 billion, representing a 28% margin and 67% year-over-year growth. Adjusted diluted EPS is expected to increase 70% to $4.02.
Block’s Q2 results tie together two increasingly important parts of its strategy. AI is being used to reduce the friction involved in developing and operating software, while proprietary transaction and customer data are being applied to products such as Moneybot, Borrow and Cash App Score that can deepen financial relationships with users.
KEY QUOTES:
“We built goose in early 2024 to work with any model because betting on a single lab means inheriting its ceiling. That led to us creating Builderbot to own orchestration across Block’s entire codebase, and in June, agentic AI helped write and review nearly all of our production code changes. Moneybot is generally available with over 1 million weekly engaged accounts and Managerbot is already automating marketing, margin analysis, and operational fixes for sellers.”
“All of it runs on one internal platform for model access, routing, tools, and permissions, with evaluation systems that let us improve quality independently of the underlying models. When a better model ships we can switch, and everything we’ve built gets better the same day.”
Jack Dorsey, Block Head, Chairman, and cofounder of Block
