According to TechCrunch, GitHub has announced the launch of GitHub Spark at its annual GitHub Universe conference in San Francisco. Spark, an experimental project from GitHub Next labs, allows users to build small web applications using natural language prompts. This tool aims to simplify the app development process by enabling users to describe their desired app in natural language, which Spark then translates into a live preview within seconds. Users can refine and iterate on their app by interacting with the bot in a chat-like interface. While experienced developers can still access and edit the underlying code, this step is optional. The core of Spark includes a GitHub repository, GitHub Actions, and Microsoft’s Azure CosmosDB as the default database for applications requiring one. GitHub CEO Thomas Dohmke emphasized that Spark is designed to facilitate personal software creation, allowing users to explore ideas and build small applications or bots to assist in their daily tasks. He noted that Spark is not intended to replace professional developers but to serve as a tool for experimentation and idea generation. There are currently no limitations on the complexity of applications that can be built with Spark, though this may change after the technical preview. Spark can utilize any web API and leverage AI models, offering users the choice between Anthropic’s Claude Sonnet and OpenAI’s GPT models. Users can share their Sparks with customized access controls and build upon shared code. Developers can also review and modify the code if needed, especially when the AI makes errors. This flexibility allows for further development and customization of applications created with Spark.