Open comments for this post
AiRY is an experimental AI-assisted programming language designed to bridge natural language and executable code. Instead of requiring strict syntax like traditional languages, users write instructions in plain, human-like language, and the system translates them into a structured intermediate representation called AiRY Core, which is then executed by a lightweight interpreter. The core idea is a two-layer architecture: a flexible natural language input layer handled by an AI model, and a deterministic execution layer that ensures consistent program behavior. So far, I have built a working Python-based interpreter for AiRY Core that supports variables, conditionals, loops, and basic arithmetic operations. On top of that, I integrated a language model pipeline using Hugging Face (google/gemma-4-31B-it) to convert natural language instructions into AiRY Core syntax using a custom prompt system. I also defined the initial specification of AiRY Core, including rules for set, show, read, if/else/elseif, and basic logical and arithmetic operations, along with a rule system for handling errors and unsupported features. The current focus is on improving translation consistency from natural language to AiRY Core, reducing ambiguity in interpretation, and stabilizing the prompt design. The long-term goal of AiRY is to make programming closer to intent rather than syntax, allowing users to describe what they want to achieve instead of how to write it in a specific language. I am very excited to make this project hope you like it!!!!
Open comments for this post
AiRY is an experimental AI-assisted programming language designed to bridge natural language and executable code. Instead of requiring strict syntax like traditional languages, users write instructions in plain, human-like language, and the system translates them into a structured intermediate representation called AiRY Core, which is then executed by a lightweight interpreter. The core idea is a two-layer architecture: a flexible natural language input layer handled by an AI model, and a deterministic execution layer that ensures consistent program behavior. So far, I have built a working Python-based interpreter for AiRY Core that supports variables, conditionals, loops, and basic arithmetic operations. On top of that, I integrated a language model pipeline using Hugging Face (google/gemma-4-31B-it) to convert natural language instructions into AiRY Core syntax using a custom prompt system. I also defined the initial specification of AiRY Core, including rules for set, show, read, if/else/elseif, and basic logical and arithmetic operations, along with a rule system for handling errors and unsupported features. The current focus is on improving translation consistency from natural language to AiRY Core, reducing ambiguity in interpretation, and stabilizing the prompt design. The long-term goal of AiRY is to make programming closer to intent rather than syntax, allowing users to describe what they want to achieve instead of how to write it in a specific language. I am very excited to make this project hope you like it!!!!
