Ah, AI. It's getting everywhere, isn't it? It has us fascinated, and sometimes maybe a little spooked.
Time for the truth: teams that don’t leverage AI will fall behind.
As a CTO (or engineering leader), part of your job is now leading this drive toward safe, thoughtful, and impactful AI adoption.
I haven’t written a comprehensive list. I don’t think anyone actually needs one. What I have done in this article is chosen the “best-in-class” tools for each area of software delivery (or at least those which currently have AI tools built for them that are capable of having an impact).
Let’s go and look at my AI for software development picks.
The AI “revolution”
It’s a cliche, but it isn’t an exaggeration. AI is revolutionizing software engineering, with a wide range of machine learning algorithms and AI-based tools and technologies emerging that can help developers improve their workflow and boost efficiency.
AI has genuinely practical applications for a wide range of tasks, from developing more accurate test cases to creating more efficient code. Developers can use AI-enabled tools to automate many of the tasks involved in software engineering, freeing up more time to focus on core development tasks. AI-based software testing is another exciting area of development. With the help of neural networks, developers can test their code more thoroughly and identify potential bugs and errors before they become significant issues.
It’s also a fast lane to improve the developer experience. AI engineering enables your team to do more of what they actually enjoy in the development process.
The future of software development is looking brighter than ever, thanks to the incredible power of AI and deep learning. With the help of these technologies, developers can streamline their workflows, improve their code quality, and stay ahead of the competition with less human intervention.
GitHub Copilot X: Your [Future] AI Pair Programmer
This one is easily the best-known on our list, and the most hyped.
The catch?
It isn’t out yet. When it arrives, it will probably be the most-adopted AI tool in software engineering.
You might be familiar with GitHub Copilot, and its features like Copilot chat. X is the smarter successor, built on GPT-4. It’s pitched as your AI pair programmer, leveled up. It integrates into most parts of your engineers’ workflows.
Here are a few things it should be able to do:
- Explain pieces of code
- Fix errors
- Generate unit tests
- Write pull request templates
- Write CLI commands
My bet is that when this arrives, it will streamline most aspects of software delivery – any part of the lifecycle directly dealing with the code.
Sourcegraph Cody – Write code faster with AI
Can’t wait for Copilot X? Try Sourcegraph Cody. It can help you read, write and understand code much faster. They say 10x faster.
Cody reads and comprehends your entire codebase along with your code graph and company docs, and can answer questions about it.
It’s in Beta (like many AI tools right now) and doesn’t always get things right, but it’s clearly a priority product for the Sourcegraph team, and they say its results get better every day.
Or try:
Mutable.ai – Another alternative. Can do autocomplete, write blocks of code, and prompt-drive development. Can’t do testing yet at the time of writing.
Codium.ai – Specialising in test-writing
Readable – Never write comments again
You want clear, consistent documentation.
Readable AI automates the process of generating comments for your source code.
Poor (or absent) docs can be the bane of engineers’ lives – it makes it especially difficult for engineers new to a codebase or a team, wasting a tonne of time.
Readable AI should significantly reduce developers' time writing comments, allowing them to concentrate on more complex tasks and improve overall productivity.
It’s compatible with the IDEs your team already uses, like VSCode, Visual Studio, IntelliJ, and PyCharm, and it can read most languages.
Adrenaline AI: Understand your code
For most engineering teams, spending what feels like a lifetime poring over Stack Overflow is an everyday reality.
What if, instead, your engineers could just ask an expert who had total comprehension of the codebase, and get an immediate answer?
That’s what Adrenaline is trying to do. It can explain how features work, locate where it’s implemented, or guide them through debugging.
It’s built using static analysis, vector search, and advanced language models.
Grit.io: Automated Technical Debt Management
What if you could automate code migrations? What if you could let AI handle dependency upgrades?
Grit.io is an automated technical debt management tool designed to do this easily. It can auto-generate pull requests that handle the nitty-gritty by adhering to the best practices, and it has a constant radar that detects regressions.
They claim migrations can be completed 10x faster. If that’s even half-true (I didn’t verify this), dealing with these kinds of tech debt will no longer be a significant drain on time and resources.
Codeball AI: Supercharge Your Code Review Process
Tired of spending countless hours reviewing pull requests? Codeball AI is here to save the day!
This AI-powered code review tool evaluates pull requests, highlights risks, and can even approve low-risk PRs with your organization’s configuration. Ultimately, it should mean you can ship faster and with greater confidence.
It’s great for spotting risky code changes – it’s been trained on millions of code contributions.
Trained on millions of code contributions, Codeball AI can recognize risky code changes and provide actionable, team-specific insights.
It works with GitHub Actions and supports 20+ programming languages.
That’s it!
Well, there you have it.
AI is changing the game. I’m absolutely convinced that this is just a mere taste of what is to come.
AI for software engineering is here. Keeping up with the change will be make or break for many tech teams, and potentially the businesses they’re a part of. Get it right, and I think it’s entirely possible to find numerous practical potential uses of AI that really can turn “10x-ing” from an anecdote into a reality.