Secretive AI coding assistant startup Augment raises $227M to rival GitHub’s Copilot – Business
Artificial intelligence-powered coding assistant startup Augment Inc. stepped into the limelight today, announcing a bumper $227 million Series B funding round that brings its valuation to $977 million, just shy of “unicorn” status. The company says its mission is to empower developers with powerful AI capabilities that will ultimately transform the way software is developed.
Today’s round was led by high profile investors including former Google LLC Chief Executive Eric Schmidt and venture capital firms such as Index Ventures, Sutter Hill Ventures and Lightspeed Partner Ventures, and marks the entrance of yet another well-funded player in the AI coding industry.
Augment’s co-founders include Igor Ostrovsky, a former chief architect at Pure Storage Inc. and software engineer at Microsoft Corp., and Guy Gur-Ari, who previously worked as an AI researcher at Google. The company is led by its Chief Executive Officer Scott Dietzen, whose resume lists Pure Storage, Yahoo! Inc. and BEA Systems Plc, a U.K.-based aerospace technology company.
Ostrovsky says software engineering is rapidly evolving to become a collaboration between human coders and AI. “At Augment, we are working with extraordinary focus to empower teams around the world with AI, enabling them to build better software,” he said.
The startup says its AI coding tools are designed to have an expert understanding of codebases and operate at the speed of thought. They’re meant to support entire teams rather than just individuals, and have features built in that will ensure the protection of intellectual property.
A veil of secrecy
Strangely though, that’s about all Augment is willing to share about its technology for now. Although the company claims to have exited stealth mode, its barebones website suggests that isn’t really the case, providing few details on what, if anything, differentiates its platform from alternative and better established AI coding tools such as GitHub Inc.’s Copilot.
In an interview with TechCrunch, Ostrovsky declined to comment on the user experience, nor was he willing to say anything about the generative AI models that power its platform, except to mention that it was fine-tuned using “industry leading” open models.
The company intends to reveal more when it launches its platform in general availability at some stage this year. It hasn’t said exactly when that will happen, but it did reveal it plans to make money from standard software-as-a-service subscriptions.
The company’s secrecy may have something to do with the highly competitive nature of the nascent industry it wants to disrupt. By far the biggest player in AI coding today is GitHub, whose Copilot tool is currently used by more than 1.3 million paying customers, including 50,000 enterprises. Other rivals include Amazon Web Services Inc.’s CodeWhisperer and Google’s Gemini Code Assist, and a host of startups such as Tabnine Ltd., Codegen Inc., Refact, Cognition AI Inc., Laredo Labs Inc., TabbyML and Magic AI Inc.
The above players are all aiming to disrupt a software industry where annual spending exceeds $1 trillion, but still fails to meet the expectations of most enterprises. Augment cites a report by Gartner Inc. that notes how software remains fragile, complex and expensive to maintain, and says developer teams are often finding themselves bogged down with long backlogs for feature requests, security patches, bug fixes, integrations and so on.
It’s believed that AI can help to fix the problems in software development, and Gartner forecasts that 50% of all enterprise software developers will be using AI-based coding tools by 2027.
The challenge for AI coders: Costs and code quality
Augment wants to be a part of the coming AI coding revolution, but it remains to be seen if there will be enough room for so many different players. One of the main problems faced by AI coding startups is the eye-watering costs involved in developing generative AI. It’s notable that GitHub reportedly loses between $20 and $80 per user with its Copilot tool due to the high cost of running the models that power it. In December 2022, the seven-year-old AI coding startup Kite.AI announced it was shutting down due to its high costs and an inability to properly monetize its offerings.
For its part, Augment says it has already generated a fair bit of momentum, claiming that hundreds of software developers from dozens of companies are using its platform in early access. One of those companies is the payments startup Keeta Inc., also backed by Eric Schmidt, which has reportedly seen a 40% increase in developer productivity since it started using Augement’s tools.
Besides having to fight tooth and nail for market share and revenue, Augment must also overcome some of the significant, industry-wide problems that continue to dog AI-generated code, such as vulnerabilities, poor quality code and copyright issues.
For example, a recent study by GitClear found that AI-powered coding assistants are pushing more poor-quality, error-strewn code into codebases, causing significant headaches for software maintainers who then have to go in and fix it manually. Moreover, there have been repeated warnings from security researchers about how generative AI-generated code is often littered with bugs and vulnerabilities.
Copyright infringement may also derail a few AI coding startups. Most generative AI coding assistants and copilots tend to be trained on publicly available data, and there are fears that may are simply reusing existing, copyrighted code. While some AI firms have argued that the fair use doctrine protects their model’s outputs from copyright infringement claims, a number of coders have filed class action lawsuits alleging intellectual property violations.
Moreover, while Gartner may be bullish on the prospects of AI coding adoption, it conceded in a 2023 report that such tools currently only provide “incremental quality-of-life improvements” and fail to deliver significant productivity gains.
Augment’s CEO Dietzen did at least concede this point, and hinted that the startup might just have the solutions enterprises are waiting for. “Current AI for code platforms are falling well short of realizing the potential of generative AI, and we are thrilled to be launching a product to close this gap,” he promised.
The size of today’s round certainly suggests Augment’s backers are confident in its approach to AI coding, and with any luck it will soon become apparent why that is. The startup said it will use the funds to accelerate product development, build new features and expand its go-to-market team ahead of its general availability later this year.