Over a Year Exploring GitHub Copilot at Itequia

Over a year ago, at Itequia, we decided to incorporate GitHub Copilot into our development tools. In a previous article, we presented the initial results of putting GitHub Copilot to the test, showing how it began to impact our development speed. Today, after more than 12 months of experience with the implementation of GitHub Copilot, we want to share a deeper analysis of how this tool has transformed our productivity, as well as the lessons learned during this time.

Integration of GitHub Copilot: A Year of Experience

Active Users and Adoption

Since the implementation of GitHub Copilot, we have had up to 8 developers with active licenses. Developers have integrated the tool into their daily workflow at different times. The adoption was gradual, allowing the team to familiarize themselves with Copilot and optimize its use for repetitive and creative tasks. Currently, we have 4 developers who consistently use Copilot in their projects. This process has allowed our professionals to maximize the potential of the tool, adapting to its different capabilities and areas of use.

Evolution of Productivity Metrics

Over time, we have observed a constant improvement in our productivity.

Below is the evolution of the metrics from March 2023 to October 2024:

How we process the data

To evaluate the impact of GitHub Copilot on our productivity, we implemented a rigorous and systematic approach to data collection and analysis. Below, we detail the steps followed to obtain the results presented in the “Evolution of Productivity Metrics” section:

• Estimated hours: Initially, we summed all the hours projected for the tasks assigned to the developers, based on previous experience and task complexity.
• Actual hours worked: Subsequently, we meticulously recorded the actual hours worked by the developers on these tasks, ensuring precision in tracking the time invested.
• Percentage of hours worked: To calculate this percentage, we divided the actual hours worked by the estimated hours and multiplied the result by 100, providing a clear measure of temporal efficiency.
• Percentage of time improvement: Finally, we subtracted the percentage of hours worked from 100. This value reflects the percentage of time improvement, indicating the efficiency gained compared to the initial estimates.

Use Cases: Practical Examples

Web Application

A web application using TypeScript. Initially, 40 hours were estimated to complete the task, but thanks to GitHub Copilot, only 15 real hours were needed.

Copilot has proven to be a valuable tool in generating new entries in the HTML from changes in the component logic.

Despite Copilot sometimes presenting unexpected behaviors, such as generating incorrect code snippets under certain conditions, it has proven especially valuable in repetitive tasks that would normally consume a lot of development time.

Mobile Application

For the mobile version of the same project, we also used TypeScript. As with the web case, 40 hours were estimated for the task, but only 15 real hours were needed.

Copilot facilitates the generation of new entries in the HTML, allowing developers to focus on styling. Its behavior can be erratic at times, but in a controlled environment, it is a great help.

Management Web System

Recently, we implemented evolutions in a management web system developed with ASP.Net technology, TypeScript, and SQL Server database.

Thanks to GitHub Copilot, we reduced programming times by 18%, decreasing from the estimated 60 hours to 49 effective hours. This 11-hour saving allowed for faster and more efficient delivery of new features.

Additionally, it allowed us to improve code quality in some critical points, optimize several important SQL Server processes, and generate more robust unit tests, which subsequently allowed us to detect errors not found in previous tests.

Using GitHub Copilot provided us with a significant advantage, allowing us to make improvements to our application more efficiently. Reducing development time by 18% shows how this tool can increase productivity and improve the quality of the final software.

Benefits of GitHub Copilot: Beyond Productivity

Increased Productivity

GitHub Copilot has been crucial in increasing our productivity. It is especially useful in repetitive tasks such as generating boilerplate code or simple snippets. This automation allows our developers to focus on higher-value tasks, such as software architecture design or solving complex problems.

Automation of repetitive tasks

Copilot has freed up a significant amount of time by automating repetitive tasks. This feature is especially useful in creating simple functions, form entries, and similar tasks. Without Copilot, these tasks consume an unnecessary amount of time. This has reduced the operational load on developers, allowing them to progress on larger projects.

Generation of Unit Tests

One of the areas where GitHub Copilot has been exceptional is in the automatic generation of unit tests. Using frameworks like Jest, Copilot has helped us set up initial tests. This allows developers to focus on more specific and complex tests that truly add value to the development process.

Suggests Classes and Methods

A significant advantage of GitHub Copilot is its ability to suggest classes and methods based on best practices. This helps maintain code quality and standardize the work of developers. It also makes it easier to integrate code among different team members.

Technical Training and Reduction of the Learning Curve

Less experienced developers have greatly benefited from Copilot. Not only does it reduce the learning curve, but it also acts as a learning companion. It provides suggestions and code examples that quickly enhance their technical skills.

Code Consistency

GitHub Copilot has played a key role in maintaining coherence and consistency in code. This is especially important in large projects with diverse teams. Its suggestions align with industry standards, making code reviews easier and ensuring the overall quality of the product.

Dynamic Knowledge Base

One of the most powerful features of GitHub Copilot is its ability to continuously learn and adapt to our codebase. As it interacts with our code, its suggestions become more personalized and accurate. This improves overall efficiency and reduces errors.

Final Reflections: What We’ve Learned and Where We’re Headed

The implementation of GitHub Copilot at Itequia has been a process of continuous improvement. While its initial focus was on enhancing productivity, we soon realized that its impact extends beyond the simple automation of repetitive tasks. Copilot has proven to be a valuable tool for the technical training of less experienced developers. By offering contextualized suggestions and code examples, it has facilitated the understanding of best practices and design patterns, significantly reducing the team’s learning curve.

As we explore more features and use cases, we’ve learned something crucial: staying competitive in software development requires combining automation with the continuous enhancement of our skills. GitHub Copilot not only optimizes technical tasks but also contributes to an ongoing process of professional growth for our developers. This learning process extends to all teams, where both less experienced and more advanced members discover new ways to optimize their code and improve process efficiency.

As we delve deeper into GitHub Copilot’s capabilities, we remain attentive to future possibilities and their impact on our projects. Our commitment to quality and innovation drives us to keep improving. We aim to deliver increasingly efficient and effective solutions to our clients.

Are you curious about how GitHub Copilot can boost your development team’s productivity and quality?

At Itequia, we’ve experienced firsthand how this tool can transform operational efficiency and foster continuous learning for developers. If you’re ready to explore how GitHub Copilot can enhance your software projects, we invite you to get in touch with us.

Itequia Team