In an era where AI is reshaping industries at breakneck speed, Doug Seven finds himself at the forefront of a technology that’s changing the way developers approach their work.
As the director and general manager of AWS AI Developer Experiences, Seven is at the helm of this transformation, where specialized AI is not just assisting but actively reshaping how code is written and deployed.
With years of experience at Microsoft and Amazon, Seven has witnessed the evolution of developer tools, but nothing compares to the potential he sees in AI agents.
“The more you trust the AI, the less you’re going to need some ‘human-in-the-loop.’ It’s not that different, really, than a lot of technology we’ve already experienced in our daily lives,” he says.
Seven’s observations on the future of coding and AI dovetail with recently published comments by AWS CEO Matt Garman and Amazon CEO Andy Jassy.
On this episode of the Shift AI Podcast, Seven sits down with host Boaz Ashkenazy to delve into how these groundbreaking digital agents are revolutionizing developer productivity, freeing engineers from mundane tasks, and unlocking new levels of efficiency. They explore what this AI-driven future means for the next generation of developers, offering a glimpse into the rapidly evolving world of software development.
Listen below, and continue reading for highlights, edited for context and clarity. Subscribe to the Shift AI Podcast and hear more episodes at ShiftAIPodcast.com.
Evolution of Developer Roles with AI
“As an individual developer, I now have one or many agents at my disposal. I’m actually a team of developers. If you think about the capacity that I have to say, ‘OK, well, actually I’m going to become the orchestrator of my team of agents,’ I’m going to have work that I’m going to take on, and then I’m going to choose what work gets dispatched to agents and what work requires me and my cognitive abilities. Maybe I’m going to have the more mundane work done by the agents and the more challenging, cognitive, creative problem-solving work done by me as a human because I can apply a different approach to it.”
The Challenge of Personifying AI as an Agent
“One of the hardest things in software development is naming things. I think I like the term ‘agents’ in the sense that it really gives you a mental model for how they work and what they do. I actually think we do naturally personify AI quite a bit because it helps us understand what it’s doing and how it’s working. And maybe some of it comes from our lifetime of watching sci-fi movies, but I think it helps build a mental model for how to interact. If I have this concept of an agent that I think of as being independent, I can actually assign work to this agent and it’s going to go do its thing. So I’m really comfortable with the name and I think it’s a great way to embody it.”
Trusting AI in High-Stakes Environments
“There’s a combination of where we are technologically and where we are cognitively with our ability and willingness to trust AI to do these things. We have to approach that very slowly. I think that’s kind of part of the ethical approach. Let’s take it at a pace that we’re comfortable with and that people can feel comfortable with and it can be successful. If I were to just come in and say, ‘Hey, I’ve got an AI that’s going to do everything for you,’ you’d probably be pretty skeptical.”
The Potential for AI to Address Real Business Needs
“We have a long history of doing Java development and we’ve applied AI to about 30,000 production Java applications. We estimate, conservatively, about 4,500 developer years saved by having AI do this work for us and then because of what we’ve been able to do by upgrading to more modern versions of the JDK and using Java, we estimate that’s about $260m in annual savings for the operations of those applications. Really remarkable in terms of what that means for us as an organization.”
Trust and the Role of Human in the Loop
“I actually believe that there’s this corollary of oppositional force between ‘human-in-the-loop’ and trust in AI. The more you trust the AI, the less you’re going to need some ‘human-in-the-loop.’ It’s not that different, really, than a lot of technology we’ve already experienced in our daily lives. The amount that you trust cruise control in your car has to do with how much you’ve used it and how familiar you’ve gotten with it, and then eventually you start to trust adaptive cruise control. Eventually, maybe you trust it to do lane changes and so on and so forth. Generative AI around software development is kind of the same thing.”
Describe the future of work in two words
“Very…Promising. I think that we are just scratching the surface of what we can do with AI. We’ve had a really seismic shift in how we can use machine learning and Generative AI to the point where what is going to happen in the next 10 years is going to be so incredible and it’s going to change how we think and how we work and how we approach problems and that to me is very promising.”