Apple is teaming up with AI startup Anthropic on a new platform for “vibe coding” using artificial intelligence. The platform would use AI to write, edit and test code on behalf of programmers.
While Apple is planning to launch the software internally, it hasn’t yet decided if it will launch publicly. This system makes use of Apple’s programming software XCode and Anthropic’s Claude Sonnet model, according to Bloomberg. This approach is similar to that of companies like.
READ: Amazon-backed Anthropic raises $3.5 billion (March 4, 2025)
While in its earlier days Apple used to be resistant to the use of generative AI for creating software for customers. However, the company seems to be changing its tune with the advancement of LLMs. Last year, Apple announced its own AI-powered coding tool for XCode called Swift Assist, intending to roll it out in 2024, however it never actually shipped to developers. Engineers have internally complained that the company’s own system could hallucinate — or make up information — and even slow down app development. The partnership with Anthropic can be seen as an acknowledgement the company needs outside help for perfecting the system. Anthropic’s Claude is known to be one of the best when it comes to programming.
This move indicates a greater willingness on Apple’s part to partner with others after struggling to develop homegrown technologies. Previously the company has avoided third parties, apart from a deal with OpenAI for the use of ChatGPT with Siri, Apple’s virtual assistant. Apple is expected to add Gemini from Alphabet Inc.’s Google as an alternative to ChatGPT later this year. The rest of the Apple Intelligence platform, which includes features like custom emoji, writing tools and organizing notifications, uses models developed by the company internally.
“Vibe coding” refers to an AI-driven programming practice in which a developer communicates the desired functionality in plain language, a concise prompt to an LLM fine-tuned for code generation. The code is then generated by the LLM, shifting the developer’s role from writing the code to manually guiding the LLM model to write the code. This term was first coined by Andrej Karpathy, co-founder of OpenAI and former Tesla AI director, in an X post.