Devika AI is an open-source AI software engineer project that helps developers program by interpreting natural-language instructions, breaking down tasks, researching necessary context, and generating code. It runs locally and supports modern LLMs such as Claude 3, GPT-4, GPT-3.5, and local models via Ollama.
Software development often involves tedious tasks like repetitive coding, planning steps, and researching tools or APIs. Devika AI aims to automate these workflows, enabling developers — from beginners to experienced engineers — to accelerate project creation, fix bugs, generate features, or scaffold code more efficiently. Being fully open-source and free increases accessibility and community contributions.
Devika runs as a local web app using Python and other dependencies (e.g., Bun, Ollama). You interact with it through a chat-style interface where you describe your objective. It plans split-step actions, conducts web search if needed, and outputs code in multiple languages. You can review or refine the generated code in real time. It’s extensible, configurable, and integrates various powerful LLM backends.
