Writers are persnickety creatures. George R.R. Martin wrote all of the Game of Thrones books using DOS and the WordStar word processor — both artifacts from the 1980s. But when it comes to fussiness about their tools, writers of words don’t hold a candle to writers of code.
Also: Brace yourself: The era of ‘citizen developers’ creating apps is here, thanks to AI
We coders are extremely particular about the coding environments and setups we use. Many of us work on codebases with hundreds of thousands to millions of lines of code. A misplaced semicolon or period might be all that stands between working code and legions of users banging down your doors with pitchforks and torches, or, even worse, a marketing guy insisting on shipping your product this quarter.
That’s where development environments (commonly referred to as IDEs) come into play. There are a whole bunch of them, and each coder has their preferences. Personally, I prefer phpStorm and VS Code, but everyone is different. But choosing an IDE isn’t the whole story. Every person’s IDE implementation is wildly customized in terms of features, window pane arrangement, shortcut keys, plugins, colors, and so much more.
When generative AI tools like ChatGPT were first released, many coders were amazed that the AIs could help us code. That was good enough. But over time, as we integrated AI into our coding workflow, it became apparent that constantly copying and pasting code snippets into various browser tabs was not only annoying — it was inefficient.
Also: 10 key reasons AI went mainstream overnight – and what happens next
So AI vendors, who used these tools themselves on a daily basis, started to integrate AI functionality right into the development environments coders use every day. You could highlight a block of code and the AI could read it, respond to it, and possibly update it. Assuming the AI didn’t screw up (which I’ve found that many AIs do when coding), the process could be a big time-saver.
All of that brings us to Google’s announcement today about Gemini Code Assist.
Free AI coding assistant for individuals
Last year, Google unveiled Gemini Code Assist. I was cautiously optimistic because some of the capabilities seemed really quite useful.
On the other hand, Gemini at the time passed only one of my four coding tests, positioning it at the bottom of the good-at-coding ladder we regularly update here on ZDNET. It’s all well and good to have an in-IDE coding interface, but only if the code produced actually works.
Also: If your AI-generated code becomes faulty, who faces the most liability exposure?
But when Gemini Code Assist was announced last year, it was only available to Gemini Advanced and other business customers. Google is announcing now that Gemini Code Assist will be available free, for everyone to use.
This is big. Among other things, the more people who use this code assist tool, the more bug reports will be sent in and the more fixes will be made. Programming tool adoption often skyrockets when that tool is available for free. Suddenly, all the nights-and-weekends programmers, along with all the student coders, gain access to the tool.
Generous usage and IDE integration
But it’s not just that the tool is now available for free. A bunch of the integrated AI development tools have free trials or limited functionality. Essentially, you can try them out, but just as soon as they prove to be useful, you hit a wall and either pay or lose out.
Also: How I test an AI chatbot’s coding ability — and you can, too
Often, the way AI providers limit access is based on the number of queries or code completions performed each month. But what does that really mean?
There’s been a long-held trope that programmers code roughly 50 lines per day. That’s net of all the changes, fixes, tool updates, and all the rest. But that’s only a trope. I know I’ve pumped out just 50 lines on some days, while other days I’ve cranked out 200 or more lines of code. Sometimes, when fueled with enough caffeine and protein, I’ve cranked out 300-400 lines.
But for our purposes, let’s say a programmer cranks out 100 or so lines of code per day. Using AI might increase that speed, so let’s assume an AI-assisted coder can do 250 lines each day. And let’s assume each of those lines represents a code completion (where the AI writes part of the code) or a prompt of some kind.
Also: The best AI for coding in 2025 (and what not to use — including DeepSeek R1)
Many AI providers let you have 2,000 code completions per month before you have to sign up for a paying plan. Using my rough estimate above, you’d run out of completions in a little over a week. That’s enough to get a taste for whether the tool might be useful, but it’s not enough to do useful work or count on it being available.
To be sure, if you’re programming as a job, you might expect to pay for the tools you use. But recreational programmers and those new to programming won’t have a budget for tools. I sure didn’t when I got started.
That’s where the next part of Google’s announcement comes in. Google is allowing 180,000 code completions per month. With that much runway, I doubt any programmer will hit a wall mid-month.
The feature is available in VS Code, JetBrains IDEs (like phpStorm), Firebase, and Android Studio.
AI-powered code reviews with GitHub
Concurrent with the announcement of the free availability and generous code completion capacity of Gemini Code Assist, Google is also announcing Gemini Code Assist for GitHub.
Also: From zero to millions? How regular people are cashing in on AI
This is a tool that works on GitHub that allows developers to perform code reviews on pull requests or check-ins. This causes Gemini to look through the code and point out problem areas that might need more attention.
Interestingly, because different coders and groups use different coding styles, the GitHub Gemini tool will allow developers to set up custom style guides for their coding styles.
Coding help for everyone
What do you think about Google making Gemini Code Assist free? Have you tried AI-powered coding tools before? If so, how do they compare? Do you see AI code reviews as a helpful addition to your workflow, or do you prefer human reviews? Will the generous code completion limit make AI a bigger part of your development process? Let us know in the comments below.
You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.