In a previous video by Eric, he mentioned a project he was very optimistic about—Augment. Today, I specifically looked into it, and the direction of this project is similar to GitHub Copilot, both focusing on developing AI programming assistants.
a strong interest. Sam Altman also mentioned in an interview that one of the future directions for the AI field is AI programming.
AI learning programming languages (such as Python) is fundamentally similar to learning natural languages (like English or Spanish). The structure of programming languages is more regular and logical, so AI may find it easier to learn programming languages than natural languages. The underlying architecture of AI is naturally suited to this rigorous, structured language learning, which also means that progress in the field of programming might be faster and more profound compared to natural language processing.
Company Background
The official website of Augment is https://www.augmentcode.com/.
Igor Ostrovsky, a former Microsoft software developer, believes that soon there will be no developers who do not use AI to assist in their workflow. In an interview with TechCrunch, he said, "Software engineering remains a difficult and often tedious and frustrating job, especially when working at scale. AI can improve software quality, boost team productivity, and help developers rediscover the joy of programming." Therefore, Ostrovsky decided to build an AI-driven coding platform that he himself would want to use. This platform, named Augment, has raised $252 million in funding, valuing it close to unicorn status (post-funding valuation of $977 million).
Former Google CEO Eric Schmidt and venture capital firms such as Index Ventures, Sutter Hill Ventures, Lightspeed Venture Partners, Innovation Endeavors, and Meritech Capital have all participated in the investment. Augment aims to disrupt the still nascent market of generative AI coding technology.
Empowering Engineers
There is an interesting blog post on the official site that shares the philosophy of Augment:
Against the backdrop of increasing fear, uncertainty, and doubt (FUD) about AI replacing all knowledge workers, including developers, Augment conveys the opposite view:
Don’t fire Kevin for Devin just yet. Augment Kevin with super powers! @dalmaer
It's easy to anthropomorphize AI tools, especially when conversing with them or influenced by depictions of AI in science fiction. The Augment team believes that systems should be built more like J.A.R.V.I.S. rather than HAL.
When developing these systems, it is crucial to remember that humans and computers each have unique strengths. True miracles happen when humans are in control, supported by ubiquitous, interconnected computer systems. In this way, not only can the personal work experience of developers be improved, but teams and organizations can achieve greater success while reducing workload and communication costs.
The Future of Software Development
Wish
I love programming. Whenever I write some code, the day usually goes well. This creative process, which produces tangible results, has a special appeal to my brain. Any platform, tool, or service that keeps me in the flow of development becomes a favorite. There is an art to breaking down an idea and gradually realizing it.
The more I can stay on the path of writing executable code, or quickly get effective help back on track when the code doesn't run, the better I feel.
Conversely, whenever I do something that feels monotonous, or I truly get stuck, my mood worsens.
Throughout my career, there have been several instances where I've seen how AI technology can provide significant assistance:
I once worked with a research team at Google X, who built models before the LLM/Transformer era to help senior software engineers manage an evolving single codebase. This process is usually very tedious and perfectly suited for computer intervention. I also worked on a project at Shopify using LLMs to help developers simplify the complexity of GraphQL, making it easier to integrate with merchant data. This experience taught me some important lessons, such as: ) demo Building a great product that works at scale in the real world is very hard A single LLM cannot solve all use cases Not just quantity... data quality is also crucial During iteration, a system capable of excellent evaluation is essential
These projects have shown me the dramatic changes that will occur in software engineering in the future, and the combination of AI technology and developers will be the driving force behind this transformation.
Currently, you can apply for the whitelist for Augment. I submitted the form and am still waiting.
