Most AI software development tools, such as GitHub Copilot, use large language models (LLMs) that are trained on huge data sets. For Copilot, this means that the model has been trained on thousands of lines of code available on GitHub. They act as a kind of “autocomplete” for code: they predict which code the developer is likely to write next and complete it automatically.
At Blis Digital, everyone, from developer to tester, has now embraced AI. Microsoft Copilot and Blis Chatty, our GPT-4o chatbot in Microsoft Teams, help us with many tasks every day. Testers, for example, use AI to make bug reports more efficient. Instead of writing long descriptions, we can now have reports generated automatically based on screenshots or videos.
The results are impressive. The biggest advantage is that AI automates repetitive tasks, so that you, as a developer, stay in your “flow” and can focus on more complex issues.
More than just writing code
Of course, there are far more AI tools than GitHub Copilot and in-house chatbots. AI helps developers make large-scale code changes, find bugs, and generate documentation. In addition, there are innovative ways to support programmers while coding. In Phind, for example, they can ask questions about their code to solve problems faster.
Tools like Builder.io and Visual Copilot bring AI to the visual design process. With these tools, non-designers such as product owners can make design adjustments without in-depth knowledge of design principles. Simple instructions such as “move the banner to the right” or “add a button” are enough to generate a usable design, significantly lowering the visual design threshold.
AI in testing and deployment
Another key area where AI has a major impact is in software testing. AI tools can now generate at least 80% of the unit tests automatically. At Blis Digital, we use GitHub Copilot for this process, which not only saves time, but improves the quality, coverage, and consistency of our tests.
In addition, we see AI innovations in the testing of user interfaces (UI). Microsoft Playwright and ZeroStep are excellent examples of how AI makes building and maintaining UI tests much faster and more efficient.
And although AI is still in its infancy in the deployment process, important steps are also being taken here.
Challenges, restrictions, future... And what it all means to you
Despite all the benefits, AI tools are still not perfect. GitHub Copilot, for example, works great for simple tasks, but the quality often leaves much to be desired for more complex tasks. It is also very important to understand that AI is not (yet) a substitute for human creativity and critical thinking. Nevertheless, much has already changed in how we make software, and more changes are coming.
Read more in our white paper
Following a recent webinar about AI in development, Christian Boer created a longread with an overview of the status quo in this area. He added his vision for the future of development, but especially what it will mean for you — our customer.
You can download that piece as a PDF here:
Read white paper: AI in Software Development
.jpg)
