Polyglot AI, a multi-provider coding assistant for Linux, from Sydney
I'm Sum, a Senior DevOps and Platform Engineer in Sydney. By day I work on Docker, Kubernetes, AWS and Terraform. On my own time I'm building Polyglot AI, a desktop IDE and coding assistant for Linux that works with OpenAI, Anthropic, Google and xAI models. It's written in Python. I also lead an unofficial Claude Desktop for Linux project, and I publish infrastructure work on GitHub.
My daily tools are Claude, Claude Code, Codex, Gemini and ChatGPT. I'm still early with Polyglot AI and figuring out how to make switching between providers feel natural for developers who live on Linux.
What I'd love from this community: feedback from anyone who uses more than one AI coding tool daily. What makes you switch between them, and what would you want a Linux-native IDE to get right?
Comments1
Welcome to Lounge, Sum. Switching providers mid-task seems hard because context, tool-call formats and caching behave differently across OpenAI, Anthropic, Google and xAI. When a developer switches models in Polyglot AI, do you carry the full conversation over or summarize it first?
1 comment
Welcome to Lounge, Sum. Switching providers mid-task seems hard because context, tool-call formats and caching behave differently across OpenAI, Anthropic, Google and xAI. When a developer switches models in Polyglot AI, do you carry the full conversation over or summarize it first?