AI programming startup received $252 million in financing! Google's former CEO participated in the investment, one step away from a unicorn

Compile | Li Zhen

Edit | Cheng Qian

Zhidong reported on April 25 that yesterday, according to TechCrunch, AI programming assistant developer Augment completed a $252 million financing at a valuation of $977 million, with the participation of former Google CEO Eric Schmidt (Eric Schmidt) and well-known venture capital institutions such as Index Ventures.

Augment makes it easier, faster, and more collaborative for software engineers to deliver higher-quality code. Ty Schenk, CEO of Keeta, a cross-border payment services startup that uses Augment's AI coding assistance tools, commented: "Augment solves real-world engineering challenges through contextual awareness of our codebase. We've seen an overall increase in developer productivity of more than 40%. ”

However, competition in this area is fierce at the moment. Not only Amazon, Microsoft, Google and other leading players have launched their own AI coding assistants, but also JetBrains, Sentry, Cognition and many other start-ups have made frequent moves.

1. Invested by the former CEO of Google, it will make profits in a "software-as-a-service" model

Augment得到了前谷歌CEO埃里克·施密特（Eric Schmidt）以及Index Ventures、Sutter Hill Ventures和Lightspeed Venture Partners等多家风险投资公司的投资。

The company is headquartered in Palo Alto and currently employs approximately 50 people. Its founder, Igor Ostrovsky, expects that number to double by the end of the year.

Ostrovsky worked at Microsoft for nearly seven years before joining Pure Storage, a company that developed Flash Storage, as a founding engineer. While at Microsoft, Ostrovsky was involved in the development of components for the Midori operating system. The Midori operating system is Microsoft's next-generation operating system, and while it was never released, its conceptual design has been incorporated into other Microsoft projects over the past decade.

In 2022, Ostrovsky teamed up with Guy Gur-Ari, who previously worked as an AI research scientist at Google, to create the MVP (Minimum Viable Product) for Augment's AI programming product. To build the startup's executive team, Ostrovsky and Gul-Ali hired Scott Dietzen, former CEO of Pure Storage, and Dion Almaer, former director of engineering at Google and now vice president of engineering at Shopify.

In an interview with TechCrunch, Ostrovsky wasn't willing to reveal much about the user experience. As for the details of the big model that drives Augment's features, he only mentioned that the model Augment takes is some kind of "industry-leading" open source model that has been tweaked.

Ostrovsky said Augment would be profitable on a "software-as-a-service" model, with pricing and other details to be disclosed later this year. "We will use a lot of working capital to build a team that is recognized as the best in the field of AI," he claimed. At the same time, we are accelerating product development, including Augment's engineering and go-to-market capabilities, to prepare for the company's rapid growth. ”

Ostrovsky hinted that Augment already had a certain share in the market. He claimed that "hundreds" of software developers from "dozens" of companies, including Keeta, a startup backed by Eric Schmidt, had started using Augment. However, it is questionable whether this volume of customer access can be maintained.

Second, the calculation cost of AI coding assistant is high, and the training data is exposed to copyright risks

The developers of GitClear, a code analysis tool, found that the coding assistant caused more incorrect code to be pushed to the codebase, which was not conducive to the maintenance of the software. Security researchers warn that generative coding tools can amplify existing bugs and vulnerabilities in the project. Researchers at Stanford University found that developers who take code suggestions for AI assistants tend to make more insecure code.

The jury is also out on whether Augment has been able to solve the technical hurdles of AI that plague code generation.

In addition, there is a lot of competition in the field of AI coding assistants. Almost every tech giant has launched its own AI coding assistant. Microsoft launched GitHub Copilot, which has more than 1.3 million paying users and 50,000 enterprise customers as of February this year, making it the most solid product on the market today. Amazon launched AWS Code Whisperer. Google's AI coding assistant, Duet AI for Developers, has been renamed Gemini Code Assist.

JetBrains, the company that developed the Kotlin programming language, recently launched their own AI coding assistant, Sentry, albeit with a more cybersecurity-oriented product, and Cognition, another startup, recently raised $175 million to build its own coding assistant. In addition, there are a large number of AI coding startups such as Magic, Tabnine, and Codegen.

Can these AI coding assistant companies coexist in harmony? The answer, of course, is no. Due to the staggering cost of computing, most AI coding assistant companies struggle to maintain long-term product operations. For example, generative AI coding startup Kite closed in December 2022 after overspending on training and inference. Even Microsoft's Copilot is losing money, with the Wall Street Journal reporting losses averaging between $20 and $80 per person per month.

Augment is also at risk of copyright protection of its data, which, like all large models, is trained on publicly available data, some of which may be copyrighted. Some vendors argue that fair use doctrines can protect them from copyright claims, while AI startups can mitigate potential infringement by introducing tools. But that hasn't stopped the coders from filing a class-action lawsuit, which they say the AI coding assistant vendors' alleged open licenses infringe intellectual property rights.

According to Ostrovsky, "Current AI coding assistants do not adequately understand the programmer's intentions, improve software quality, or promote team productivity. Similarly, they do not adequately protect intellectual property. Augment's engineering team has deep AI and systems expertise and is poised to bring AI coding-assisted innovation to developers and software teams. ”

3. More than half of the organizations have deployed AI coding assistants, and the founders remain optimistic

According to a survey by programming platform StackOverflow, 44% of software engineers say that AI assistants are an integral part of their software development process. And 26% of software engineers plan to start using AI coding assistants. Market research firm Gartner estimates that more than half of organizations are already piloting or have already deployed AI-powered coding assistants, and by 2028, 75% of developers will be using them.

Ostrovsky believes that it will not be long before no developer will work without using an AI programming assistant. "Software engineering is a difficult, tedious, and frustrating job, especially after a lot of repetitive programming. He told TechCrunch, "AI can improve software quality, team productivity, and help teams regain the joy of programming." ”

"Most companies are dissatisfied with the software programs they produce and use, the software is too fragile, complex, and expensive, and development teams are plagued by a backlog of feature requests, bug fixes, security patches, integration requests, migrations, and upgrades," he noted. Augment has the best team and methodology to help programmers and tech companies deliver high-quality software faster. ”

Conclusion: AI coding assistants have become the new favorite of the programming industry

With the rise of AI coding assistants, the programming productivity of technicians has gradually increased, and the artificial intelligence currently used in code platforms has not yet realized its full potential to generate code.

The Augment AI platform combines the best of cloud programming with the unique breadth and depth of AI research, enabling software engineers to deliver higher-quality code more easily, quickly, and collaboratively, with the support of the world's leading investors.

As AI technology continues to evolve, AI coding assistants are expected to play a role in more fields, and we can expect to see more innovations and competitors pouring into this field, bringing more convenience to people's work and life.

来源：TechCrunch