Is This the End of Software Engineering As We Know It? Meet Devin: The AI Dream (Or Nightmare?)

Heard of Cognition ’s Devin: the AI software engineer and haven’t had the time to dig deeper? Here is a 30 sec tldr; of what’s happening:

• Devin is an autonomous AI software developer introduced by Cognition AI, with significant funding from Peter Thiel and others.?
• Designed to independently code or assist engineers, standing out with its ability to plan and execute tasks.
• Demonstrates a 14% success rate in bug fixes, outperforming ChatGPT's 2%.
• Its market readiness remains uncertain, with potential yet to be fully revealed.
• If successful, Product Managers/ non coders will no longer need software engineers to build software, fix bugs etc. (of course - I joke)
• Watch the demo if you’re interested here. ?

Intrigued? Uncover my exclusive insights on this potentially industry-changing? development in just 3 minutes.

What is Devin?

Devin is a new "concept" AI software engineer created by Cognition Labs, designed to code and assist software engineers and more importantly “work independently”.?

Which means if you are not a software developer - you could assign tasks to Devin, such as creating a website or app that maps Italian restaurants in Istanbul, and the software will autonomously perform necessary searches, gather data like addresses and contact information, then build and publish a website/app with this information. Throughout the process, Devin displays the tasks it's executing and autonomously finds and fixes bugs.?

Bloomberg journalist Ashlee Vance, who had an opportunity to test Devin, reported that it could build a website within 5 to 10 minutes. Read more details of his experience here?

Who is behind Devin?

A stealth startup, Cognition AI, only recently established as a corporation two months ago, is led by Scott Wu , its chief executive officer; Steven Hao , the chief technology officer; and Walden Yan , the chief product officer. It has raised $21 million from Peter Thiel’s Founders Fund and other prominent investors like Elad Gil.?

Why is it interesting?

Devin stands out for its planning capability, a significant advancement over large language models like ChatGPT. While ChatGPT has a basic understanding of language and can code to some extent, it often loses context and struggles with multi-step tasks, achieving only a 2% success rate in fixing coding bugs. Devin, however, showcases a 14% success rate in similar tasks, marking it seven times more effective. Cognition Labs highlights these capabilities, though it has yet to fully release Devin.

Is it ready for prime time now?

Hard to say till it's more open and folks get their hands on it but the vision is definitely exciting with the possibilities this presents. Cognition AI also doesn’t share a lot about the technology’s underpinnings other than that the founding team found unique ways to combine large language models (LLMs) such as OpenAI’s GPT-4 with reinforcement learning techniques. Given its a fairly infant company, I personally would be surprised if it is really ready to be a disruptor just yet.?

Who else is doing this?

Magic AI is another noteworthy startup pursuing the development of AI coding tools. Magic AI recently raised over $100 million to create a "superhuman software engineer." Unlike some firms that rely on existing large language models (LLMs) from companies like OpenAI or Anthropic,?

Why do startups announce ambitious goals before their technology is fully developed?

Startups often present visionary, seemingly impossible goals to captivate attention, secure investments, and stir public and industry interest, even before their technologies are fully developed.?

This strategy, demonstrated by companies like SpaceX, which announced its unthinkable goal of making space travel more affordable and ultimately colonising Mars, helps secure early funding and partnerships critical for research, development, and scaling. By publicly committing to ambitious objectives, these startups not only position themselves as pioneers but also build the momentum needed to overcome technical and financial hurdles, gradually turning their once-impossible visions into reality.?

This approach has proven successful in driving innovation, establishing market leadership, and fundamentally changing industries.

Should we be worried about AI replacing our jobs?

My hot take:

Do we need to be worried about AI right this minute? Probably NOT.??

Do we need to be paranoid and prepare for what work will look in the next 5 years? YES WITHOUT A DOUBT!

It's important for folks, software engineers and others, to pay close attention to the ongoing innovations in AI technology.

?? Being proactive and reskilling is required as we transition into a future where the dynamics between humans and machines are expected to undergo a significant transformation.

Currently, we are accustomed to a world where machines and technology primarily serve as tools to assist humans in various tasks, enhancing our productivity and capabilities.

However, the rapid pace of technological advancements suggests that we are on the cusp of a major paradigm shift. Within the next decade, it's plausible that the roles between humans and machines will evolve, with machines taking on more autonomous, decision-making functions.?

Are you reskilling for AI, if yes I would love to hear how, comment and let me know what you are doing to stay ahead of the curve?

Read more about people management in the era of AI here.

Learned something new reading this newsletter? Subscribe to get fresh new takes on the latest developments in the product management and AI space every 2 weeks (funny memes included)!