AI is greatly enhancing coding, and developers are embracing it.
In a recent StackOverflow poll, 44% of software engineers said they currently use AI tools as part of their development process, and 26% said they plan to use them in the near future. Gartner estimates that more than half of organizations are currently piloting or have deployed AI-driven coding assistants, and that 75% of developers will use some form of coding assistant by 2028. .
Igor Ostrovsky, a former Microsoft software developer, believes that soon there will be no developer who does not use AI in their workflows. “Software engineering remains a difficult, often tedious and frustrating job, especially at scale,” he told TechCrunch. “AI can help improve software quality, team productivity, and bring back the joy of programming.”
So Ostrovsky decided to build an AI-powered coding platform that he himself would like to use.
That platform is Augment, which emerged from stealth on Wednesday by raising $252 million in funding at a post-money valuation close to unicorn ($977 million). With investment from former Google CEO Eric Schmidt and VCs including Index Ventures, Sutter Hill Ventures, Lightspeed Venture Partners, Innovation Endeavors, and Meritech Capital, Augment is building a market that is still in its infancy for generative AI coding technology. We aim to shake things up.
“Most companies are dissatisfied with the programs they produce and use. Software is often fragile, complex, and expensive to maintain, and development teams are often dissatisfied with feature requests, bug fixes, and security patches. , stuck with a long backlog of integration requests, migrations, and upgrades,” Ostrovsky said. “Augment has both the best team and recipe to help programmers and their organizations deliver high-quality software faster.”
Mr. Ostrovsky spent nearly seven years at Microsoft before joining Pure Storage, a startup that develops flash data storage hardware and software products, as a founding engineer. While at Microsoft, Ostrovsky worked on components for Midori, a next-generation operating system that the company never released, but the concepts have been incorporated into other Microsoft projects over the past decade. I did.
In 2022, Ostrovsky and Gai Gar Ali, previously an AI research scientist at Google, collaborated to create Augment's MVP. Ostrovsky and Gul Ali will hire former Pure Storage CEO Scott Dietzen and former Google director of engineering and Shopify VP of engineering to fill the startup's executive positions. We welcomed Mr. Dion Armour, who served as the
Augmentation remains strangely secretive.
In our conversation, Ostrovsky didn't want to say much about the user experience or even the generative AI model that powers Augment's functionality (whatever that is). Except Augment uses some kind of fine-tuned “industry-leading” open model.
He said how Augment plans to make money: standard Software-as-a-Service subscriptions. Pricing and other details will be revealed later this year, Ostrovsky added, closer to the Augment's general availability date.
“Our funding provides a long-term runway to continue building what we believe is the best team in enterprise AI,” he said. “As the company prepares for rapid growth, we are accelerating product development and strengthening Augment's product, engineering and go-to-market capabilities.”
Rapid growth is perhaps Augment's best bet at making waves in an increasingly cutthroat industry.
Virtually every tech giant offers their own version of an AI coding assistant. Microsoft has his GitHub Copilot, which has the strongest foundation ever, with over 1.3 million individual customers and 50,000 business customers as of February. Amazon has CodeWhisperer for AWS. And Google has Gemini Code Assist, which was recently rebranded from Duet AI for Developers.
There are a number of other coding assistant startups, including Magic, Tabnine, Codegen, Refact, TabbyML, Sweet, Laredo, and Cognition (which reportedly raised $175 million). Harness and JetBrains, the creators of the Kotlin programming language, recently released their own programming language. Sentry was similar (albeit more cybersecurity-oriented).
Will all of them, and now Augment included, be able to do business harmoniously? It's unlikely. Eye-popping computing costs alone make the AI coding assistant business a difficult business to sustain. Overruns related to training and model delivery forced generative AI coding startup Kite to shut down in December 2022. According to the Wall Street Journal, even Copilot loses anywhere from about $20 per month to about $80 per user per month.
Ostrovsky hinted that Augment already has momentum, with “hundreds” of software developers at “dozens” of companies including payments startup Keeta (also backed by Eric Schmidt). Claims to be using Augment in early access. But will the intake last? Indeed, that's the million dollar question.
I also wonder if Augment has taken any steps toward resolving the technical issues plaguing code-generating AI, particularly regarding vulnerabilities.
An analysis by GitClear, the developer of the code analysis tool of the same name, found that coding assistants are increasingly pushing incorrect code into codebases, causing headaches for software administrators. Security researchers have warned that generative coding tools can amplify existing bugs and exploits in projects. And researchers at Stanford University found that developers who accept code recommendations from AI assistants tend to write less secure code.
The next thing to worry about is copyright.
Augment's model, like all generative AI models, was definitely trained on publicly available data. Some of them may be copyrighted or under restrictive licenses. Some vendors argue that the fair use doctrine protects them from claims of copyright infringement while also allowing them to deploy tools to mitigate potential infringement. But that hasn't stopped programmers from filing class-action lawsuits over open licenses and claims of intellectual property infringement.
To all this, Mr. Ostrovsky says: “Current AI coding assistants do not properly understand programmer intent, do not improve software quality, do not promote team productivity, and do not adequately protect intellectual property. Augment Engineering The team boasts deep AI and systems expertise, and we are ready to bring AI coding assistance innovation to developers and software teams.”
Palo Alto-based Augment currently has approximately 50 employees. Ostrovsky expects that number to double by the end of the year.