The Impact of GenAI Tools on Your Development Process

Unless you have been living under a rock for the past two years, chances are you have encountered one or more Generative Artificial Intelligence (GenAI) tools for software development. Today, many developers use GenAI tools to enhance their development processes and workflows.

In this blog post, we will discuss GenAI tools for software development and how they can impact your development process.

Since the introduction of the first public version of ChatGPT in late 2022, the number of GenAI tools on the market has increased exponentially. ChatGPT is an example of a generalized GenAI model that is suitable for a vast number of different tasks, yet it is not specialized for any single one. There are other types of models as well, such as image generation models.

Specialized GenAI tools are available for software development. These tools are still based on other generalized models but trained specifically to have skills relevant to software development. This includes models that are experts at writing source code, that specialize in writing test suites, and more.

How models are trained and how they are specialized for software development is part of a larger topic of the AI development life cycle. This is outside the scope of this blog post.

Today, many developers enjoy the benefits of using GenAI tools as part of their development workflow.

A GenAI tool should support your software development workflows without being an obstacle or slowing you down.

GenAI tools are generally based on large language models (LLMs), which are trained on large data sets. In the context of software development, these datasets consist of source code written in various programming languages. By training an LLM on source code, it will be able to suggest code based on the prompt that you give it.

When you interact with a GenAI tool such as ChatGPT through a browser or a mobile app, you supply it with explicit prompts directly as text or voice input. However, the prompts you supply for GenAI models in software development are generally implicit.

For example, GitHub Copilot is a GenAI software development tool that is integrated into IDEs, such as VS Code. While it is possible to interact with these tools through a chat interface with explicit prompts, the most common way to interact with them is implicitly as you write your code; these tools will suggest code based on the contents of the files that you have opened in your editor and the current code you are typing.

Below, we see an example of working with GitHub Copilot, where it suggests a completion of the code we began to write. It bases its suggestion on the content of files we have currently opened in our editor.

If you want to use the model for a focused request, you can use different commands. For example, GitHub Copilot supports commands such as /doc, /explain, /fix, and others.

There are many benefits to using GenAI tools in development. In the following sections, we will cover a few of the most essential benefits that GenAI tools bring to your development workflow.

Using a GenAI tool does not guarantee the benefits listed below, but when you learn to utilize the tools fully, you will reap the rewards.

Once you get used to working with a GenAI tool, you will increase the speed at which you write source code. These tools can generate whole classes and functions or just help you complete the line of code you are currently writing.

In the example below, we use GitHub Copilot to generate source code for an HTTP client in TypeScript from a single inline comment. The comment is used to describe what you want. You can use a simple description or a more thorough description to get more specific code suggestions.

Writing documentation for your source code is important for several reasons, primarily to educate other developers and users on how the source code works. However, documentation is often outdated shortly after it is written.

Keeping your documentation up to date is difficult, but with the help of GenAI tools, you can quickly generate documentation of your source code. In the following example, we use GitHub Copilot to generate a docstring for a selected TypeScript function:

You can also use the GenAI tool to interactively ask questions about your code, which you would normally turn to the documentation to find the answer for. This allows even junior developers to quickly understand the source code.

In the example below, we ask GitHub Copilot to explain a selected piece of code. The prompt can be customized to ask for specific details that would be difficult to predict when writing normal documentation.

Writing secure code is important, yet it is something that many junior developers and even a few senior developers have little or no experience with. GenAI tools can help you write more secure code, including security best practices, vulnerability detection, and more. The GenAI tool can also review code you have written and provide feedback and suggestions on how to improve it from a security standpoint.

Unfortunately, writing tests for your source code is often overlooked, postponed, or skipped altogether. With the help of GenAI, you can quickly generate whole test suites that cover the main code flows and edge cases you have not thought about.

The following image shows an example of using GitHub Copilot to get started with testing. Here, Copilot notices that we have not set up testing for this project yet, but we can use an interactive guide to get help.

If you have an idea about a software project you want to pursue, you can utilize GenAI tools to quickly prototype an idea to determine its viability. Even if the generated code does not follow your normal coding practices, you can still evaluate the idea’s feasibility before spending precious development hours on it.

A GenAI tool can understand existing codebases, and you can use this knowledge to your advantage. When you onboard new developers to a project, they can use this knowledge to quickly learn about the codebase.

In the example below, we use the /explain command with GitHub Copilot to get a detailed explanation of the selected code and understand what it is doing.

GenAI tools can help you develop code with fewer errors. GenAI models are trained on large samples of source code with the collective knowledge of many developers over a long time. This knowledge is now available using GenAI models.

When using a GenAI tool for software development, you should evaluate the tools and models you use to make sure the code it produces fulfills your organization’s coding practices.

Even if the GenAI tool generates code, you will likely have to go through the code with a critical eye, just like you would review code from other team members.

You should monitor the use of all the GenAI tools you onboard to your organization to understand their impact. This will help you understand how each GenAI tool enhances different key metrics and give you a more detailed view of how accurate the suggestions each tool provides are.

To make sure your developers use the GenAI tools you have to their fullest potential, you should arrange team training and establish guidelines.

The number of GenAI software development tools will likely increase in the coming years. Specialized software development models purposely trained for specific use cases will be the norm. Generalized models will still have a purpose in software development, but they will never be able to compete with specialized models.

We will likely see GenAI tools integrated into many tools for software development, similar to what we have seen for VS Code and JetBrains IDEs in the past few years.

The biggest impact of GenAI tools so far has been in the implementation and testing phases of the software development life cycle (SDLC). However, we will likely see more focus on GenAI in other SDLC tools going forward.

We have already seen GenAI tools that you can assign tickets to and have them implement a whole feature in your code base. It might still be a few years before we can fully trust a GenAI model to work autonomously as part of a development team in our organization. However, this is where development is likely.

GenAI tools for developers are more important than ever. Although they will not completely replace developers anytime soon, they play an important role in the developer experience and lead to improved development outcomes.

A good GenAI tool can help you increase your coding speed, write better and more secure code, help you generate test suites, and more. In the future, we will likely see GenAI models that can be a part of our development teams in a more autonomous way.