This article describes the birth, development and impact of the AI coding assistant Devin. Starting from a "magic operation" that solved the Christmas server problem, Devin quickly became a programmer's right-hand assistant and was even adopted by companies such as Microsoft. Its emergence not only aroused the industry's attention to the AI coding revolution, but also made programmers around the world think about the future of their profession. This article will take an in-depth look at the team behind Devin, its technical advantages, market prospects, and its potential impact on programmers’ careers.
Remember the server problem that left a group of Silicon Valley elites scratching their heads just before Christmas last year? At that time, the Cognition AI team was struggling with a complex data server. They tried various methods but still couldn't get it to work properly. Just when everyone was about to give up, someone suggested that the AI coding assistant Devin, which had not yet been officially unveiled at the time, try it. Guess what? Devin actually solved the problem like magic! It just deleted one For a system test file that was ignored by the team, the server terminal indicator light changed from red to green. It was a miracle!
Devin's "magic operation" shocked the entire team, and they realized that the field of software engineering was about to usher in a revolutionary change. A year later, Devin has become a useful assistant for coders in their daily work. It can find and fix bugs, update code blocks, migrate code between different platforms, and even "clean up this code base" according to simple instructions. ” From the Lord make plans and execute them. Unlike tools like GitHub and Codeium that only provide code suggestions, Devin is an autonomous agent that is theoretically capable of writing, fixing, and migrating code independently without human involvement, and can complete the entire project typically assigned to developers.
Behind Devin is a "dream team" composed of a group of talented programmers - the founding team of Cognition AI. This team can be called an IOI gold medal "harvester". The three founders, Scott Wu, Steven Hao and Walden Yan, are all IOI gold medalists. It is said that they have won 10 IOI gold medals in total! Scott Wu, a serial entrepreneur, Before founding Cognition AI, he created an AI-driven social network platform called Lunchclub. He has been a mathematical genius since he was a child, and has achieved the highest level of "Legendary Master" ranking on Codeforces (a programmer exchange website). Steven Hao has rich experience in AI systems. He graduated from MIT with a major in computer science and mathematics and worked as a senior AI system engineer at Scale AI. Walden Yan is a young and promising Harvard graduate. This team of talented teenagers built Devin in just 6 months and received US$176 million in investment from Founders Fund and Khosla Ventures. The company's valuation soared to US$2 billion!
Well-known companies such as Ramp, MongoDB and Microsoft have begun to use Devin. Microsoft CTO Kevin Scott even praised Devin as an extraordinary tool at the annual developer conference. Devin isn't perfect, however. A YouTube blogger with 35 years of experience as a software engineer reproduced a promotional video of Devin completing an Upwork task, only to find that Devin's actual performance was far less impressive than shown in the video. Some people also question that Cognition AI exaggerates Devin's capabilities, believing that Devin can currently only complete some predefined tasks, such as cleaning up existing code.
Although Devin still has shortcomings, its development potential is still huge. Cognition AI is constantly improving Devin's capabilities, such as allowing it to mobilize subordinate AIs to assist work, which will form an "AI bureaucracy" like a "legion" composed of junior engineers. This "management" model may make some programmers uneasy, but Scott Wu believes that Devin can help companies launch more projects and allow human programmers to focus on more meaningful work.
AI-generated code is already starting to reshape entire industries. Google CEO Pichai said that more than 25% of new code within Google is written by AI. Microsoft CEO Nadella said that GitHub's code auto-completion tool has accounted for 40% of this year's revenue growth. AI coding has become the most funded use case in the GenAI field. In the first half of 2024, startups in this field raised more than US$1 billion. It is expected that by 2029, the actual revenue of AI coding will exceed US$4 billion.
The future of AI coding is exciting, but also dangerous for millions of programmers around the world. Does the emergence of Devin mean that the programmer profession is about to disappear? Scott Wu believes that large-scale layoffs are unlikely to occur because the current programmer market is still in short supply. But no matter what, the wave of AI coding has swept across. Programmers, are you ready?
The rapid development of AI coding technology has brought impacts and challenges to the programmer industry, but it has also brought new opportunities. Programmers need to adapt to new technological trends and improve their skills in order to remain invincible in future competition. Programmers in the future may focus more on the development, application and management of AI tools, as well as more creative and strategic work.