Thoughtworks, an international tech consulting company, released the latest edition of their Thoughtworks Technology Radar report, which aims to track and rate various software trends affecting the industry. An advisory board of directors meet and discuss current topics to advise software developers on which projects to pursue or pass, using the metrics, “Adopt”, “Trial”, “Assess”, and “Hold”. The latest report gives strong recommendations towards PyTorch for deep learning while suggesting that developers assess other machine learning and AI-powered tools and their risks.

“Adopt” is a strong suggestion that industry professionals should use the technology in their projects. “Trial” refers to tech trends that are worth using, so long as teams consider the potential risks and downsides. On the other hand, trends given the “Assess” ranking are only worth exploring in certain use cases and should be heavily researched, whereas “Hold” is a cautionary indictment.

Machine learning

It comes as no surprise that artificial intelligence was a heavily discussed topic in the latest report. In this field, Thoughtworks offers the following assessments:

• Adopt PyTorch: The machine learning framework is used for a large range of deep learning use cases, including natural language processing and image classification. According to Thoughtworks’ advisory board, PyTorch excels beyond TensorFlow and offers a wide variety of State-of-the-Art models.
• Assess AI-powered development tools: With advancements in machine learning tools, developers are reconsidering how they work and finding new ways to improve their productivity. However, before developers go all in on using machine learning to assist with their code, they should assess if the approach will work for them and ensure that legally, they are in the clear to use the intellectual property in these datasets.
• Assess prompt engineering: Thoughtworks notes that while using prompt engineering for model outputs, there could potentially be some security risks, and the domain is not yet fully understood. As the conversation progresses around AI, software teams should pay attention to this domain.
• Assess ChatGPT: Naturally, there are many concerns as well as legal and ethical questions surrounding ChatGPT. Here, Thoughtworks recommends that all potential ChatGPT users discuss the tool and its uses with their legal teams before proceeding.
• Assess GitHub Copilot: GitHub’s artificial intelligence-powered coding assistant can help developers in their coding tasks, through IDE integration, code generation, and code predictions. However, it can introduce some bugs.

Programming languages & frameworks

New frameworks and libraries often receive a lot of initial hype and fanfare but can be subject to heavy criticism for not delivering on their promises. Here is what developers should be aware of, according to Thoughtworks:

• Adopt Gradle Kotlin DSL: Teams that are still using Groovy are recommended that they begin migrating away from the language and towards Kotlin instead. Here, Gradle Kotlin DSL will help assist when creating new projects with Gradle.
• Assess Qwik: The new JavaScript framework focuses on exactly what it sounds like: quick loading times in the browser. It does so without hydration techniques by using resumability and greatly reducing the amount of JavaScript code that needs to be loaded. When we spoke with Misko Hevery, Chief Technology Officer at Builder.io, about it, he said: “The value is not focus on performance, but rather focus on what you think is important and Qwik will ensure that the startup is instant. Qwik takes the performance problem out of the developer focus.”
• Assess .NET MAUI: Microsoft’s new framework improves upon Xamarin. Forms and allows developers to create native mobile and desktop apps with C# and XAML. Thoughtworks notes that teams should wait and see how .NET MAUI adoption is handled in the industry before jumping in.
• Assess .NET 7 AOT: The new capabilities of deploying ahead-of-time in .NET 7 should be researched before used in production, but can also help boost start-up times. Assess Turborepo: Turborepo is a new build system used for very large JavaScript and TypeScript codebases that aims to reduce the large overhead incurred by monorepos.

More tools and techniques to adopt

Notable highlights from the report that Thoughtworks recommends teams to adopt include:

• CI/CD as a service: Teams are strongly recommended to consider the benefits of implementing GitHub Actions, Azure DevOps, and/or GitLab CI/CD into their Continuous Integration/Continuous Deployment workflow.
• Contentful: The headless content management system Contentful comes with many advantages, including CMS as code implementation and a newly introduced app framework.
• DVC: DVC (Data Version Control) is an open-source tool used for data science projects. It can be used to train and test a data set and according to Thoughtworks, their teams have had much success using DVC while bootstrapping their projects.
• K3s: Thoughtworks highly recommends this Kubernetes distribution when used for edge computing and situations where resources are highly-constrained and a lightweight solution is necessary. It also now supports WebAssembly workloads.

Have a look at the full report here and learn more about which technology trends your team should look into, and which you should evaluate further before putting into production.