On Sunday, GitHub chief Thomas Dohmke teased upcoming announcements with four cryptic photos on X, sparking excitement among developers for what’s next. One of the images featured a firecracker lighting up the sky, hinting at GitHub’s new launch, Spark, which allows developers of all skill levels to build apps in natural language and bring their ideas to life.
These micro apps, known as ‘Sparks,’ are fully functional and can integrate AI features.
“For too long, there has been an unscalable barrier to entry separating a vast majority of the world’s population from building software. This can change with GitHub Spark, our new AI-native tool to build applications entirely in natural language,” said Dohmke.
He added that with Spark, over one billion personal computer and mobile phone users will be able to build and share their own micro apps directly on GitHub. Clearly, Dohmke wants everyone to become a developer.
Dohmke showcased Spark by building a tic-tac-toe game featuring a rubber duck and a hippo, all with just a single line of prompt.
Interestingly, the company revealed that there are over 17 million developers in India building on GitHub, representing a 28% increase in 2024, making India the fastest-growing developer community in the world.
Inspired by Claude Artifacts?
At first, Spark appears to be inspired by Claude Artifacts, which lets users build mobile-friendly and responsive applications using natural language prompts. Notably, GitHub Copilot has also partnered with Anthropic to make Claude 3.5 Sonnet available on GitHub, making it multi-model.
Previously, AIM experimented with Artifacts and successfully created a Cricket Quiz game, Temple Run, and Flappy Bird, all with a single line of prompts in English.
Similar to Artifacts, Spark helps developers visualise their projects. According to the company, users start with an initial prompt using both OpenAI and Anthropic models. They can see live previews of their app as it’s built, explore different options for each request, and automatically save versions of their work to compare as they progress.
Moreover, experienced developers can directly edit the underlying code, while novice developers can create using natural language—it’s up to them. When users are satisfied with their Spark, they can easily run it on their desktop, tablet, or mobile device, getting instant value from their creation. They can also share their Sparks with customised access controls, allowing others to remix and build on their work.
Artifacts ushered in the era of on-demand software. When using mobile phones, we often search for apps that meet our specific needs. For example, if you’re into fitness, you might download an app that offers various workouts. However, that app may not provide the customisation you seek. Now, instead of relying on downloads, you can create personalised apps that cater to your needs.
On the internet, one can find several apps built using Claude Artifacts, such as the Rubik’s Cube Simulator, Self-Playing Snake Game, Reddit Thread Analyzer, Drum Pad, and Daily Calorie Expenditure.
GitHub Spark is not Alone. Recently, Replit launched Replit Agents, which allows developers to build software using natural language prompts. It simplifies software development, making it more accessible to users of varying skill levels. Currently, the agent is only available in Repls created via the Replit Agent entry and does not support existing Repls or imported repositories.
Meanwhile, OpenAI recently launched canvas, a new interface for working with ChatGPT on writing and coding projects. It provides developers with a menu of quick-action shortcuts, such as adjusting writing length, debugging code, and performing other tasks.
Similar to Claude Artifacts, canvas’ interface allows users to work on writing and coding projects side by side with ChatGPT, offering real-time edits, feedback, and suggestions. It is integrated with GPT-4o and can be manually selected in the model picker.
“ChatGPT’s new canvas interface is a game changer. I just used it to create a tesseract/hypercube visualiser with Three.js. I’m loving the unified UX—chat, inline comments, and watching GPT-4o work its magic on the code—all in one place. It never gets old,” posted a user on X.
Many developers on the internet have been experimenting with OpenAI’s o1 and Claude Sonnet 3.5 to build fully functional apps.
“Just combined @OpenAI o1 and Cursor Composer to create an iOS app in under 10 minutes! O1 Mini kicked off the project (01 was taking too long to think), then I switched to o1 to finish the details,” posted Ammaar Reshi, the head of design at Eleven Labs.
“The OpenAI o1 model builds a fully functional chess game that lets me compete against an AI-based opponent,” shared another user on X, adding that o1-preview is the real deal.
End of Low-Code and No-Code Platforms
GitHub Spark is going to be a huge threat to low-code/no-code app builder platforms such as AppMySite, Builder.ai, Flutter, and React Native. At the same time, tools like Spark and Artifacts also allow non-techies to solve word problems simply by thinking creatively.
“Barrier to entry has become so low for building software products. There is an ocean of new stuff coming in every day. But the fundamentals hold true, building the right thing for the right people is hard and is human,” said Shane Neubauer, senior growth strategist at Prisma.
For now, it would be safe to say that LC/NC platforms have to adapt to the changes and integrate generative AI features. Jinen Dedhia, the founder of DronaHQ, said on Hacker News that he believes LLMs can eliminate the need for LC/NC platforms, and the only way to stay relevant is to tightly integrate AI capabilities into these platforms.
However, some opine that no code and generative AI can coexist.
“GenAI can dramatically speed up the development process. It can generate code for common functionalities, allowing developers to focus on the more complex aspects of their applications. For low-code and no-code platforms, this means faster app creation,” said Vidhya Radhakrishnan Chandrika, technology architect at Infosys.
She added that while low-code platforms offer pre-built components, GenAI can generate highly customised code based on specific requirements. However, if tools like Spark get much better over time, people are likely to shift away from no-code and low-code platforms.
The post GitHub Spark Spells Trouble for No-Code Platforms appeared first on AIM.
