In the December/January 2025 issue of Forbes Magazine, the story of 28-year-old Scott Wu and his team at Cognition, a San Francisco-based AI startup, is featured. With $200 million in funding, they are developing an AI tool called Devin that can program entirely on its own, like an “army of junior engineers.”
Devin’s first major task was setting up a complex data server, which it accomplished by running commands that baffled its creators. This success marked the beginning of AI taking on coding tasks such as spotting bugs, fixing code, and migrating between platforms. While other players in the field provide digital assistants for coding, Devin stands out as an autonomous AI agent that can write code itself, completing projects without human involvement.
AI-generated code is already starting to change the industry, with Google reporting that over a quarter of its new code is written by AI. The demand for AI coding tools has led to startups focused on this area raising over $1 billion in funding in the first half of 2024. Cognition has attracted numerous customers, including companies like Ramp and MongoDB, which use Devin for tasks such as writing tests, cleaning up code, and updating architecture.
Despite the potential for job displacement, investors have shown great interest in Cognition. Founders Fund and Khosla Ventures invested $176 million in a Series B round, boosting Cognition’s valuation to $2 billion just six months after its founding. The company’s founders, all with accolades in competitive coding, have a head start in developing the ultimate coding machine. Cognition’s partnership with Microsoft to offer Devin to developers on its Azure cloud further solidifies its position in the market.
While some have questioned Devin’s capabilities and whether it is living up to the hype, Cognition continues to improve the tool based on feedback from early customers. In some instances, engineers with access to Devin have been able to work eight times faster. However, there is still room for improvement, as seen in experiments where Devin launched subordinate AIs that created a never-ending loop of AI bureaucracy. Despite the challenges, Cognition sees Devin as a tool that can enable companies to pursue more projects and hire humans for more meaningful work.
As the AI coding landscape continues to evolve, the conversation around the impact on software engineering jobs grows. Wu acknowledges the fear and questions surrounding AI’s role in the industry but remains optimistic about the potential for Devin to drive innovation and create new opportunities for both AI and human collaboration. Cognition’s journey with Devin represents a shift towards a new paradigm in software engineering, one that combines the capabilities of AI with human expertise to drive progress in the field.
