Highlights:

  • Magic goes beyond GitHub’s Copilot by letting programmers express their desires in English.
  • The business also intends to utilize AI to add elements to code depending on developer requests, define what a piece of code does by simply looking at it, and even identify, forewarn about, and fix possible errors in the code.

Magic AI Inc., a startup developing an AI platform to generate code, recently announced that it has raised USD 23 million in new funding to expand its staff and get its product ready for launch.

The startup raised a total of USD 28 million in the Series A fundraising round, headed by Alphabet’s CapitalG with participation from Nat Friedman, Elad Gil, and Amplify Partners.

Software engineers can use Magic’s tool, which is currently unavailable, to express what they need in natural language as it generates code or what the business refers to as “software that builds software.” In this way, the developer and the AI can cooperate in locating code, evaluating its use, reusing it, and then discussing code changes.

The notion is that the AI would function effectively like a “colleague inside the computer” that functions as an intelligent agent and can help write and edit the code.

Many of today’s automated code generation solutions use simpler interfaces that remove the monotony of writing code by recommending code based on what developers are presently writing.

In this way, Magic is comparable to Microsoft’s GitHub Copilot, which automates software code writing using an AI model based on GPT-3. When programmers write code, Copilot offers code that automatically finishes the task and can finish even lengthy portions of code.

Magic goes beyond that by letting programmers express their desires in English. That places it in a different context than making recommendations while programmers write code.

Eric Steinberger, the CEO of Magic, stated to a well-known digital media house, “Magic aims to drastically reduce the time and financial cost of developing software. Giving teams access to an AI colleague who can understand legacy code and help new developers navigate it will enable companies to scale the impact of their current employees and train new employees with less personal coaching. In turn, employees will grow their skills faster and will be able to move among high-impact projects with increased agility.”

Magic created its neural network architecture with the ability to rapidly read code to accomplish this quickly. According to the business, it “can read 100 times more lines of code than Transformers.” An AI Transformer is a kind of machine learning model that was first created by Google researchers and is mostly utilized in computer vision and natural language processing. Standards frameworks like TensorFlow and PyTorch have integrated transformers.

According to the company, Magic’s AI code production platform will produce code and follow developers’ requests. The business also intends to utilize AI for adding elements to code depending on developer requests, define what a piece of code does by simply looking at it, and even identify, forewarn about, and fix possible errors in the code.

Developers can send messages to Magic, an AI they can converse with like a colleague, informing it what modifications they wish to be made to the code. The AI will take into account their feedback, and as the dialogue continues, the code will be adjusted to meet the specifics of the exchange.

The business stated that it plans to use the financing to expand its team from its size of six employees to 25 individuals, concentrating on engineering and development, product, and go-to-market initiatives. The service is not yet accessible, but it will be shortly. On the website, there is a signup form for interested individuals.