2 min Devops

Magic wants to compete with Github Copilot and just got funded

Magic wants to compete with Github Copilot and just got funded

Magic raised $23 million from Alphabet’s CapitalG and other investors like Elad Gil, Nat Friedman, and Amplify Partners. Magic’s CEO and co-founder, Eric Steinberger, is a former AI researcher at Meta. Magic wants to offer a code-generating platform to help software engineers write code.

Elad Gil and his co-founder, Sebastian De Ro, founded Magic to develop a code-generating platform that uses AI to help software engineers write, review, debug, and plan code changes. Magic operates like a pair programmer, allowing developers to communicate and collaborate with the tool in natural language. The tool can understand the context of coding projects and developers and continuously learn more over time.

Magic’s goal is to reduce the time and financial cost of software development

Magic can do this by providing teams with an AI colleague that can understand legacy code and help new developers navigate it, enabling companies to scale their employees’ impact and train new employees with less coaching.

However, Magic faces competition from GitHub’s Copilot, which has already been used by over 1.2 million people and has a significant following. Steinberger promises that Magic will be able to do more than Copilot, thanks to a new neural network architecture that can read 100 times more lines of code than the popular Transformer architecture.

The use of AI-powered code-generating systems, such as Magic and Copilot, raises legal questions, as they were trained on publicly available code, some of which is copyrighted.

Microsoft, GitHub, and OpenAI are currently being sued in a class action lawsuit for violating copyright law by allowing Copilot to use licensed code without providing credit. Magic is taking steps to prevent copyrighted code from appearing in its suggestions and citing the source of suggested code where possible.

Despite these challenges, Steinberger is optimistic about the future of Magic and its potential to revolutionize software development.

He admitted that training state-of-the-art models remains expensive, raising the bar for new entrants like Magic, but added that the ultimate goal is for AI to complete large tasks end-to-end without human supervision.