Your AI Coding Buddy Is Always Available at 2 a.m.
Ever had a brilliant coding inspiration at 2 a.m. but wished you had someone to bounce ideas off of? Aja Hammerly has been there. And now she believes AI is becoming that always-available coding partner developers dream of.
AI-assisted software development, Hammerly says, is like having “a pair [programmer] who’s available whenever I want.”
“If I have a crazy idea at 2 a.m. and want to go hack something together to show the team, I can sit down and just do it because I have the AI there to help me with the pieces I’m not as strong at,” said Hammerly, director of developer relations at Google, in this On the Road episode of The New Stack Makers, recorded at Google Cloud Next in Las Vegas.
Hammerly and episode host Alex Williams, TNS founder and publisher, explored the differences between DevEx AI and DevEx for AI, why Hammerly advises developers to start with AI-assisted coding before exploring AI agents and the launch of Firebase Studio, an AI-enhanced integrated developer platform (IDE) that allows developers to choose when to use AI features, and when to work manually.
Understanding DevEx AI’s Dual Nature
In the world according to Hammerly, developer experience for AI actually encompasses two distinct concepts that often get confused.
“When people talk about DevEx for AI or DevEx AI, there’s one of two things they usually mean,” Hammerly said. “The first one is, how do we as developers use AI in our applications — what tools exist for that, and how do we use those effectively?
“And the second is, how do developers use AI to eliminate toil, to make ourselves more efficient? Both are super cool, both are super exciting, and there’s a lot happening in both of those spaces right now.”
In this context, the first type of DevEx AI encompasses AI-powered coding assistants, Hammerly said. “Treat it as a pair. Someone to answer questions, to ask you questions that clarify your thinking, and give feedback.”
This virtual pair programmer experience is why she believes anyone just starting to adopt AI into their developer workflow should begin here, with AI-generated code and code completion.
“One of my favorite things to do is ask AI to explain code to me,” Hammerly continued. “When I’m learning a new codebase and someone says, ‘Hey, can you check out this code and help me debug this?’ I ask Gemini to explain the code, and Gemini gives me a very quick overview and some context. After that, I can start digging in and use my engineering skills to diagnose and debug the problem.”
If the first aspect is AI as a programming partner for writing your code, the second is AI as a problem solver for actually implementing it. In other words, Hammerly said, applying agentic AI to the parts of your software delivery lifecycle that are toilsome, difficult or just hard for your team to stay on top of.
“So much of the developer’s cognitive load is performance, it’s quality, it’s testing,” she said. “These are all places where AI can help automate the stuff that’s boring.”
In Hammerly’s case, that’s user-facing development. “I’m not great at frontend — I will own that,” she acknowledged. With AI assistance, though, she can focus on what she loves: “I can have the AI help me with the frontend stuff, remind me of how to do things, generate some of the code for me. Then I get to hack away on the stuff I like and that interests me in the backend, like business logic and hooking it up to various APIs.”
Enter Firebase Studio
Now, with the launch of Firebase Studio during Google Cloud Next, Hammerly said developers can have both kinds of DevEx AI in one convenient solution.
“Firebase Studio is an IDE with a bunch of really cool AI features,” she said. “It’s got a prototyping tool — aka ‘vibe coding.’ It’s got AI coding assistance built in, and an AI agent for automating stuff, too — but you get to choose when you use the AI, and what you use it for.”
The choice is critical, she said, because there are many different ways to write code. “We’re building this new reality where AI is part of developer workflows, and people need to find how it works for them,” she said. “Every developer I’ve ever met has a slightly different workflow. Me, I’m big on pairing. I’m big on lots and lots and lots of tests. I like test-driven development. But that’s just one way to write code. There are many other ways to write code that are also very effective.”
Hammerly’s biggest advice for developers dabbling with AI is to experiment. Figure out where AI gives you the most value and use those aspects. If something that someone else is doing doesn’t work for you, OK, don’t use it like that.
“The art of development is very individual,” she said. “At the end of the day, it’s between the person and the keyboard.”
Check out the full episode to hear more of the conversation, including why traditional monitoring practices still apply to AI apps, version control for AI prompts and why large language models are basically databases.
