Coinbase has significantly expanded its use of artificial intelligence in software development, with the company revealing that AI now assists in writing between 95% and 100% of its code.
The update highlights how rapidly AI has become integrated into Coinbase’s engineering operations as the crypto exchange continues reshaping its workforce and product development strategy. The company says AI is accelerating software delivery while allowing engineering teams to focus more on reviewing, refining, and making strategic decisions rather than writing code manually.
Rob Witoff, Coinbase’s Head of Platform, shared that nearly all code produced within the company is now generated with assistance from large language models (LLMs). Engineers typically work alongside five to ten AI agents simultaneously, using them to complete coding tasks that previously required much larger development teams.
According to the company, the collective output of these AI agents is already comparable to the work of roughly 1,200 employees. Coinbase also expects AI adoption to continue expanding, with long-term ambitions of scaling the number of AI agents supporting its engineering teams even further.
The latest figures represent a sharp increase from earlier this year. In February, AI reportedly contributed to around 40% of Coinbase’s code generation. Within a matter of months, that figure has climbed to nearly the company’s entire software output, illustrating how quickly AI-assisted development has become part of Coinbase’s day-to-day operations.
The company’s growing reliance on AI follows broader organizational changes announced earlier this year. In May, Coinbase reduced its workforce by approximately 14%, affecting around 700 employees. CEO Brian Armstrong told staff that advances in artificial intelligence had fundamentally changed how work is performed across the business and that Coinbase intended to operate with a faster, startup-style culture centered on AI-powered productivity.
Rather than replacing engineers entirely, Coinbase says developers now spend more time supervising AI-generated code, validating outputs, improving system architecture, and solving complex engineering problems. AI handles much of the repetitive implementation work, while human engineers remain responsible for reviewing quality, security, compliance, and overall product direction.
The company’s approach reflects a wider trend across the technology sector, where generative AI is increasingly being used to assist software development. Large language models are capable of generating code, identifying bugs, writing documentation, and automating repetitive programming tasks, allowing engineering teams to build and iterate products more quickly.
For Coinbase, AI adoption also aligns with its broader strategy of improving operational efficiency while continuing to expand its crypto ecosystem. The exchange has spent the past several years investing in new products across trading, payments, custody, blockchain infrastructure, and institutional services, making software development speed an increasingly important competitive advantage.
Despite the productivity gains, Coinbase has emphasized that human oversight remains essential. AI-generated code continues to undergo review before deployment, particularly for systems involving customer assets, financial infrastructure, and regulatory compliance. Company executives have stated that while AI excels at execution, strategic judgment and decision-making continue to rely on experienced engineers.
The rapid increase in AI-generated code also reflects changing expectations within the software industry. As coding assistants become more capable, developers are increasingly shifting from writing every line of code themselves to directing AI systems, reviewing outputs, and orchestrating multiple AI agents during the development process.
Coinbase’s latest disclosure provides one of the clearest examples yet of how generative AI is transforming software engineering at scale. As AI tools continue improving, companies across both the technology and cryptocurrency sectors are expected to further integrate automation into their development workflows while maintaining human oversight for critical systems and business decisions.
