Everything we do at Magic comes down to a singular goal: build aligned and complete AI to accelerate humanity’s progress on the world’s most challenging problems.

Following a $5M Seed round last summer, we’re excited to announce that we’ve raised a $23M Series A from CapitalG (Alphabet’s Independent Growth Fund), Nat Friedman (former CEO, GitHub & co-creator of Github Copilot), Elad Gil, Amplify Partners, Noam Brown (Research Scientist, Meta), Mehdi Ghissassi (Director of Product, Google Research), Bryan Pellegrino (LayerZero), Arthur Breitman (Tezos), 10x Founders, William Tunstall-Pedoe (Unlikely AI), Adam Jafer & Fredrik Hjelm (Voi), Klaudius Kalcher & Roland Boubela (MostlyAI), Xavier Sarras, Florian Huber, Daniel Dippold, and others.

This funding will accelerate our journey to build a true AI colleague for software engineering. We couldn’t imagine a better roster of long-term backers for Magic.

Historically, automation has been a net good for our society. It has led us from subsistence farming to becoming a globally connected society. We went from lifting rocks to operating cranes, from planting seeds to driving tractors, and from manual calculation to programming. As a result, humanity is richer, safer, and healthier. In the past 100 years, we have more than doubled the average life expectancy.

Much has changed over the arc of human history, but there has been one consistent trend: technological tools and automation push people to increasingly complex work. But once AI surpasses us in general intelligence, this trend will likely end – and we believe that’s good.

Today, society associates our worth with our labor. Historically, this has been beneficial. In a world of extreme scarcity, broader productivity gains map well to broader gains in well-being. But once AGI does all the work, we can separate the two. We could build a world where people are valued for their character and not for what they can produce. Perhaps, AGI can be the foundation of an economy and society of people, not workers.

We’d like to live in such a world. And we’re working to build it.

Building AGI is a hard to define endeavor. We wanted a concrete path to help us prioritize what to work on. Aiming to create a product that’s impossible to build with current technologies helps us do so – science via product. We use user feedback as our benchmark, and measure progress in product iteration.

Naturally, Magic is vertically integrated – we do research, train foundation models, build products, and talk to users. Owning the front-end gives us direct feedback; training our own foundation models allows us to incorporate it. Moreover, we believe strongly in algorithmic innovation. LLMs have shown their strength, but there are fundamental components they lack – scaling plays a role, but we’re more interested in understanding how we can improve the underlying algorithms.

We’re starting with an AI software engineer. Going from idea to code requires decision-making based on an ambiguous description, a deep abstract understanding of all context, creative and reliable engineering, and an iterative feedback loop with the outside world. While today’s code completion tools are already proving to be immensely useful, we can’t wait to see what people build with Magic.

2023 will undoubtedly be an exciting year on this journey. We’ve been handed the unique opportunity to contribute to the biggest technological revolution in a century, and we can’t do it without an amazing team.

If you’re excited by “science via product”, hard engineering problems, and user empathy, we’d love to hear your story. Get in touch at join@magic.dev

If you’re excited to have an AI colleague when it’s ready, join our waitlist.